(navigation image)
Home American Libraries | Canadian Libraries | Universal Library | Community Texts | Project Gutenberg | Children's Library | Biodiversity Heritage Library | Additional Collections
Search: Advanced Search
Anonymous User (login or join us)
Upload
See other formats

Full text of "Velocity-selective imaging using cesium-based photon detectors"

VELOCITY-SELECTIVE IMAGING USING CESIUM-BASED PHOTON 

DETECTORS 



By 

NATHAN CHARLES PLXLEY 



A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL 

OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT 

OF THE REQUIREMENTS FOR THE DEGREE OF 

DOCTOR OF PHILOSOPHY 

UNIVERSITY OF FLORIDA 

2002 



ACKNOWLEDGMENTS 

I would first like to thank my graduate research advisor, Jim Winefordner, for his 
support and friendship during the past four years. Jim provides an environment that 
supports research freedom, innovation, independent thinking, teamwork, and personal 
development. I will always remember Jim's philosophies on research, teaching, and 
leadership and aspire to incorporate these ideologies into my personality. 

I would also like to thank Ben Smith for his friendship and contributions to my 
research. Ben has given me practical advice on numerous occasions to approach a given 
problem or develop an idea. Most importantly, Ben's appreciation of science, history, 
culture, and life has helped me realize that the world is full of exciting and rewarding 
experiences and opportunities, some of which are related to hard-work and career 
development, while many are not. I leave graduate school with fond memories of the 
hours spent in Ben's office. 

Nicolo Omenetto has also played a large role in my scientific and personal 
development. His insight and knowledge extend into a variety of fields, and his 
contributions to my research are numerous. His enthusiasm and excitement toward 
science are unequaled, and even the most rigorous of scientific discussions were always 
enjoyable. 

I would also like to thank Dimitri Pappas for his friendship and contributions to 
my research. This research project would not have been possible without Dimitri's 
determination, hard work, and insight. Tiffany Correll has also made many contributions 



< • . 

in 






to this research. She has added new perspectives and personality to this project while 
also playing an important role in my personal life. Oleg Matveev and Igor Gornushkin 
have provided me with advice on numerous occasions, and I am thankful to have had 
such amazing postdocs in our laboratory. 

I would also like to thank the supporting staff, especially Jeanne Karably and 
Sonia Coleman, for efficiently handling a variety of needs ranging from purchase orders 
to travel reimbursements. My appreciation also goes to the Chemistry Department's 
Electronic and Machine Shops for good advice and top-notch work. 

I would also like to thank the past and present members of the Winefordner and 
Harrison groups for their friendship and support. I would like to specifically 
acknowledge Eric Oxley and Paige Eagan. Eric has been a great friend since our first 
days at the University of Florida. I could not even begin to list the numerous ways in 
which he has provided support. 

Perhaps most of all, I would like to thank all my friends and family for their 
encouragement, love, and support. 



iv 



TABLE OF CONTENTS 

page 

ACKNOWLEDGEMENTS iii 

ABSTRACT vii 

CHAPTER 

1 INTRODUCTION TO SPECTRALLY SELECTIVE IMAGING 1 

2 DEVELOMENT OF RESONANCE IONIZATION AND FLUORESCENCE 
DETECTION SCHEMES 10 

Introduction 10 

Two-Step Ionization Scheme 12 

Experimental Setup 12 

Results 15 

Three-Step Ionization Scheme 19 

Experimental Setup 21 

Results 23 

Fluorescence Detection Scheme 28 

Calibration of Photomultiplier Tube 30 

Measurement of Resonance Fluorescence 31 

Results 33 

Conclusions 33 

3 THEORETICAL EVALUATION OF A CESIUM RESONANCE 
FLUORESCENCE DETECTION SCHEME 35 

Introduction 35 

Methodology for Rate Equation Model 36 

Cesium Excitation Scheme 36 

Rate Equation Model of Eight Atomic Levels 38 

Results 40 

Density Matrix Simulations 44 

Conclusions 47 



4 SUB-DOPPLER SPECTRAL RESOLUTION FOR A CESIUM RESONANCE 
FLUORESCENCE DETECTOR AND IMAGING MONOCHROMATOR 49 

Introduction 49 

Photon Detection With Sub-Doppler Spectral Resolution 5 1 

Experimental Setup 52 

Results 54 

Imaging with Sub-Doppler Spectral Resolution 58 

Experimental Setup 59 

Results 62 

Conclusions 67 

5 MOVING OBJECT DETECTION AND IMAGING 69 

Introduction to Doppler Velocimetry 69 

Moving Object Detection Using a Resonance Fluorescence Detector 70 

Experimental Setup 70 

Results 73 

Imaging Doppler-shifted Photons 76 

Experimental Setup 76 

Results 79 

Conclusions 86 

6 CONCLUSIONS 88 

7 FUTURE WORK 91 

APPENDIX 95 

REFERENCES 101 

BIOGRAPHICAL SKETCH 105 



VI 



Abstract of Dissertation Presented to the Graduate School 

of the University of Florida in Partial Fulfillment of the 

Requirements for the Degree of Doctor of Philosophy 

VELOCITY-SELECTIVE IMAGING USING CESIUM-BASED PHOTON 

DETECTORS 

By 

Nathan Charles Pixley 
August, 2002 

Chair: Dr. James D. Winefordner 
Department: Chemistry 

In the field of imaging science, there is a growing trend towards the development 
of image detectors providing superior spectral resolution. Applications such as laser 
Doppler velocimetry demand spectrally resolved photon detection on the order of MHz or 
GHz. One way to achieve such spectral resolution is to utilize the narrow absorption 
features of atoms in the gas phase. Such absorption is inherently selective, and the 
absorption process can be monitored through detecting the resulting fluorescence or 
ionization due to laser excitation coupled to the excited atomic states. Photon detectors 
and imaging systems based on this concept have been previously demonstrated using 
both mercury and cesium atoms. 

The goal of this work is to further develop cesium based image detectors and 
demonstrate their spectral-resolving power through the detection of moving objects. To 
achieve these goals, early studies focused on the experimental evaluation of schemes for 
cesium ionization and fluorescence detection. Fluorescence detection was further 



vn 



investigated theoretically using rate-equations and the density matrix formalism in terms 
of quantum efficiency and identification of competing fluorescence pathways. 

Methods were devised to improve the spectral resolution below the Doppler- 
broadened absorption linewidth. In the context of Doppler velocimetry, such narrowing 
can allow for improved detection of more slowly moving objects. The improvements 
were achieved by exciting a selective population of atoms. The sub-Doppler spectral 
resolution (200 MHz) was demonstrated in both non-imaging and imaging modes of 
operation. Efforts also focused on applying these cesium-based detectors to moving 
object detection. In this last study, a rotating disc was illuminated by the first excitation 
laser, and the corresponding Doppler shift was detected or imaged. 



vin 



CHAPTER 1 
INTRODUCTION TO SPECTRALLY SELECTIVE IMAGING 

In the field of imaging science, there is a growing trend towards the development 
of image detectors that can provide superior spectral resolution. Applications such as 
laser Doppler velocimetry, chemical imaging, and long-range optical telecommunications 
demand spectrally-resolved photon detection on the order of MHz or GHz. Photon 
detection with a high spectral resolution is capable using traditional devices, and several 
imaging systems can provide a high resolving power [1-4]; however, these detectors 
often lack other critical figures of merit such as high throughput, sensitivity, spatial 
resolution, and image quality. For example, a conventional spectrometer can be used to 
obtain one- and two-dimensional image information with excellent spectral resolution. 
This resolving power is achieved, however, with a large loss of throughput. Such 
spectrometers also generate line-scanned images as opposed to two-dimensional imaging 
in a single acquisition. 

Image detectors based on the resonance absorption of photons by an atomic vapor 
can provide true, two-dimensional imaging with a spectral resolution and sensitivity 
limited by the characteristics and spectroscopy of the atomic vapor. These detectors 
require a defined excitation or ionization scheme with a specific atomic transition 
(typically from the ground state) determining the detection wavelength. Once a photon is 
absorbed, subsequent laser excitation can generate wavelength-shifted fluorescence or an 
ion/electron pair. In the case of excitation followed by fluorescence, the fluorescence can 
be detected optically and serves as the analytical signal for the initial absorption event. 

1 



When the atoms are ionized, the ionization signal can be detected electrically. An 
important characteristic of either process is that a signal (fluorescence or ionization) 
occurs only when a photon is absorbed in the first transition. When this is the case, the 
spectral response of the photon detector is defined by the absorption linewidth of this first 
transition (typically Doppler-broadened in a low-pressure cell). These selective photon 
detectors based on the observation of ionization or fluorescence are called resonance 
ionization detectors (RIDs) and resonance fluorescence monochromators (RFMs), 
respectively. The RID and RFM can be applied to imaging by expanding the excitation 
and/or ionization lasers into two dimensions. When operating in imaging mode, these 
detectors are called resonance ionization image detectors (RIIDs) and resonance 
fluorescence imaging monochromators (RFIMs). Figure 1-1 shows a general schematic 
of the RIID operation, and a schematic of an RFIM is shown in Figure 1-2. 

An inherent property of the RFIM and RIID is the narrow spectral window of 
photon detection. Applications that would most benefit from these devices are those that 
generate small wavelength shifts upon interaction with incident light. As shown in Figure 
1-3, the incident light can be wavelength tuned, such that upon interaction with an object, 
the scattered light will be in resonance with the first transition of the atomic vapor. The 
rejection of the incident light by the detector will depend on the magnitude of the shift 
compared to the spectral resolution of the photon detector. 

A RIID based on mercury was first described by Matveev et al. [5]. A three-step 
ionization scheme was used with the first transition at 253.7 nm (the detection 
wavelength). The electrons created in the ionization process were directed toward a 
luminescent screen by a voltage applied between the screen and the front window. The 



<■ 



Detector 
Surface 



K+\ 



Atomic 

Vapor 

Cell 




Phosphor 
Screen 



Metal 
Film 



Figure 1-1. Schematic of RIID operation. When photons in the vicinity of X { enter the 
RIID, only photons in resonance with the first transition are absorbed (solid arrows). 
Photons outside the narrow absorption band are not absorbed (dashed arrows). Lasers X 2 
and X3 are expanded into sheets of light and directed into the side of the cell. Atoms 
absorbing Xi, X 2 , and X 3 are ionized. Electrons are then accelerated toward the phosphor 
screen by a voltage applied between the phosphor and a metal film on the input window. 
The phosphor emits photons, creating a two-dimensional image representative of the X-i 
photons initially absorbed. This optical image can be detected using a CCD camera. 



Filter 






■\ 



Detector 
Surface 



Atomic 
Vapor Cell 



Figure 1-2. Schematic of RFIM operation. When photons in the vicinity of X\ enter the 
RIID, only photons in resonance with the first transition are absorbed (solid arrows). 
Photons outside the narrow absorption band (dashed arrows) pass through the cell 
unabsorbed. Atoms excited by Xi can be further excited by X 2 which is expanded into a 
sheet of light and directed into the side of the cell. Atoms absorbing A,i and X 2 can 
fluoresce (k fL ). This fluorescence can be detected at the back of the cell using a CCD 
camera. Due to the large difference in wavelengths between X. FL and X-i, a standard optical 
filter can be used to spectrally separate X-fl from the unabsorbed photons passing through 
the cell. 



Laser 
Source 




RIID or 
RFIM 



X+AX 



shift 



Object 



x. 



Figure 1-3. Schematic of photon detection using an atomic vapor based photon detector. 
The RFIM or RIID detects photons at A,]. A wavelength shift (e.g. a Doppler or Raman 
shift) can be detected if the incident light is tuned away from A-i by the shift of interest (ki 

± AA, S hift ). 



optical signal created by the phosphor was captured by a CCD camera. The mercury 
RIID was further improved by the addition of a microchannel plate [6]. A sealed, 
compact mercury RIID was also reported [7] where an image composed of 1000 photons 
was detected by image summation with a signal-to-noise ratio of 17; the spectral 
resolution obtained using this RIID was on the order of 25 GHz (the combination of the 
Doppler-broadened profiles of seven mercury isotopes in the RIID). 

The mercury RIID is unique in that it is the only ionization-based image detector 
based on the resonance absorption of photons by an atomic vapor; however, four different 
image detectors based on resonance absorption and fluorescence detection have been 
described previously. An RFIM based on mercury was described by Finkelstein et al. 
[8,9]. A one-step scheme was used in which the resonance absorption and resonance 
fluorescence occurred at the same transition. Since absorption and fluorescence occurred 
at the same wavelength, such a scheme does not provide for ideal stray-light rejection; 
however, specificity was achieved by temporal control of the laser pulse and detection 
system. A spectral resolution of 25 GHz and a spatial resolution of 1 mm were reported. 
An RFIM based on mercury has also been described by Matveev et al. [10]. The 
excitation scheme was two-step followed by fluorescence to an intermediate state. Two 
excitation steps followed by wavelength-shifted fluorescence detection allowed for 
fluorescence detection that was over 100 nm away from either excitation laser. A spectral 
resolution of 25 GHz was reported, and the spatial resolution was 200 urn. 

A cesium based RFIM has been described by Korevaar et al. [1 1]. This RFIM was 
based on a two-step excitation scheme followed by wavelength-shifted fluorescence. The 
detection wavelength (852.12 nm) corresponded to the 6 2 Si /2 -» 6 2 ?° m transition. Excited 



atoms in the P-state were then further excited to the 8 2 Si/2 state through laser excitation at 
794.39 nra. The detection (fluorescence) wavelength was 455.35 nm (7 2 P°3/2 -^ 6 2 Si/2) 
and 459.32 nm (7 2 P°i/2 -> 6 2 Si/2). A spatial resolution of 500 urn was reported; the 
images acquired were made up of 256 pixels. The reported spectral resolution was 600 
MHz, which is the Doppler-broadened absorption profile at the chosen operational 
temperature (100 °C). The description of the spectral resolution is complicated by the 
splitting of the ground state level into two hyperfine levels separated by 9 GHz. The 
result is two absorption channels, each 600 MHz wide and separated by 9 GHz. The 
effects of such a noise channel could be benign, depending on the proposed application. 

The most recently developed RFIM has been described by Pappas et al. [12,13]. 
This RFIM is based on two-step excitation in a cesium vapor. Just as in the Korevaar 
scheme, the detection wavelength corresponded to the 6 2 Si/2 -> 6 2 P°3/2 transition. The 
two schemes differ in the second transition; in this latter scheme, atoms in the P-state are 
further excited to the 6 2 D 5/2 state through laser excitation at 917.23 nm. The detection 
(fluorescence) wavelength was 455.35 nm (7 2 P° 3/ 2 -> 6 2 Si/2>. The spatial resolution 
achieved was better than 200 urn and the spectral resolution was limited by the Doppler- 
broadened linewidth (400 MHz) of the first transition at room temperature. Unlike the 
RFIM reported by Korevaar, the splitting of the ground state level did not result in two 
different absorption channels. 

The use of cesium as the active-element for an RFIM has several advantages. The 
ground state transition, used by Korevaar and Pappas, has a large oscillator strength (f = 
0.72); both transitions used in the latter scheme occur in the near infrared which is 
beneficial in several ways. First, wavelengths in the near infrared are accessible using 



8 

external cavity diode lasers, which are spectrally narrow, tunable, inexpensive, and 
potentially portable. In addition, these wavelengths can effectively penetrate biological 
tissues, which enables the biological and medical fields as potential applications for this 
technology, for example, the in vivo Raman spectroscopy of human tissues. Another 
advantage of cesium is its large number density in the gas phase at room temperature. In 
addition, cesium has only one stable isotope, which prevents any spectral shifts or 
apparent broadening in the atomic lines due to isotopic effects. 

The goal of this work was to further develop cesium-based image detectors and 
demonstrate their spectral-resolving power through the detection of moving objects. To 
achieve these goals, early studies focused on the experimental evaluation of schemes for 
cesium ionization and fluorescence detection. Two ionization schemes were studied, as 
well as the fluorescence scheme utilized by Pappas et al. [12,13]. The ionization schemes 
were investigated to evaluate the feasibility of a cesium RIID; the RFM scheme was 
studied to determine the quantum efficiency of the previously described cesium RFM and 
RFIM. In addition, theoretical studies were performed based on rate-equations and the 
density matrix formalism. The model developed was used to evaluate the quantum 
efficiency of the cesium RFM and also to identify processes that compete with resonance 
fluorescence detection. 

Methods were devised to improve the spectral resolution of non-imaging and 
imaging detection previously limited by the absorption linewidth. Such narrowing can 
allow for more accurate detection of slower moving objects. The improvements achieved 
by excitation of a selective population of atoms resulted in a sub-Doppler spectral 
resolution on the order of 200 MHz. Demonstration of these improvements are described 






using the cesium RFM and RFIM since the improvements could be readily implemented 
in this previously developed technology; however, these methods could be applied to any 
multi-step resonance fluorescence or ionization scheme. Additional efforts focused on 
demonstrating the use of the improved RFM and RFIM for moving object detection. In 
this last study, a rotating disc was illuminated by the first excitation laser, and the 
corresponding Doppler shift was detected or imaged using the cesium RFM or RFIM, 
respectively. In the last section of this work, methods are discussed that can potentially 
improve the spectral resolution and sensitivity of cesium-based photon detection and 
imaging. 



CHAPTER 2 
DEVELOMENT OF RESONANCE IONIZATION AND FLUORESCENCE 

DETECTION SCHEMES 

Introduction 

The performance of resonance fluorescence and ionization detectors is ultimately 
limited by characteristics of the chosen excitation or ionization scheme. The sensitivity, 
quantum efficiency, and spatial resolution (in the case of imaging) are largely determined 
by the absorption, emission, and ionization processes occurring within the atomic vapor. 
The properties of the atomic vapor cell can also influence these figures of merit. In both 
fluorescence and ionization detection, the cell must be constructed from materials that are 
non-reactive with the atoms within the cell, and the material should be transparent to 
wavelengths of either the input (excitation or ionization) or output (in the case of 
fluorescence). In fluorescence detection, these two requirements are readily met; 
however, developing an ionization cell may be problematic depending on the active 
element. In the RIID described previously [5-7, 14], mercury served as the active element 
within the cell. Since mercury is not a particularly reactive metal, no problems were 
encountered regarding reactivity between mercury and any of the ionization cell materials 
(quartz, thin-film metallic coatings, phosphors, electrical feedthroughs, etc.); however, 
several concerns must be addressed in order to successfully develop an RIID based on 
cesium. Cesium is an element with reactivity similar to sodium. Cesium is a strong 
reducer and generates explosive reactions with water. In the case of mercury, early 
versions of the RIID were implemented somewhat easily. Once a cell was constructed, a 



10 



11 

drop of mercury was placed inside the cell before sealing and pumping the cell down to 
the desired pressure. During early studies in construction of a cesium RID/RIID in-house, 
reactivity was observed with a variety of materials such as copper gaskets and teflon. 
After these initial attempts, a cesium RID was constructed by a commercial vendor 
(Opthos Instruments, Rockville, MD, USA). Opthos Instruments is a manufacturer of 
cylindrical, atomic vapor cells made of borosilicate glass. This vendor constructed a 
similar cell containing electrodes and electrical feedthroughs. This cell could not be used 
for imaging, but the evaluation of ionization schemes could be accomplished with this 
cesium ionization cell. 

The emphasis of this chapter is the evaluation of two cesium ionization schemes 
and one excitation/fluorescence scheme for the potential use in cesium RIIDs and RFIMs. 
The first scheme discussed is a two-step ionization scheme based on absorption of 
photons at the detection wavelength (852.12 nm, 6 2 Si /2 -> 6 2 P° 3/ 2) followed by direct 
photoionization by an argon ion laser operating at 496 nm. In the second scheme, a 
second transition at 917.23 nm (6 2 P 3 / 2 -> 6 2 D 5/2 ) is added before the ionizing step. 
Atoms in the D-state are then photoionized by a Nd:YAG laser at 1064 nm. The 
resonance fluorescence scheme consists of two-step excitation (6 2 Si/ 2 -> 6 2 P° 3/2 -> 
6 2 D 5/2 ) followed by radiative decay to the ground state (6 2 D 5/2 -» 7 2 P° 3/2 -> 6 2 Si /2 ). The 
detection wavelength of this scheme is 455.53 nm (7 2 P° 3/2 -> 6 2 Si /2 ). 

The quantum efficiency of all three schemes has been evaluated. The efficiency 
for a RID can be expressed as the ratio between the number of charges detected and the 
number of photons within the detection bandwidth that are absorbed by the atomic vapor. 



12 

Determining the quantum efficiency for an RFM can be done in a similar fashion and can 
be expressed as the ratio of the number of fluorescing to the number of absorbed photons. 

Two-Step Ionization Scheme 

The first transition of the two-step ionization scheme occurs at 852.12 nm (6 S1/2 
-> 6 2 P°3/2). Excited atoms are subsequently photoionized into the ionization continuum 
by an argon ion laser that can operate at several wavelengths between 457 nm and 515 
nm. The ideal wavelength for ionization corresponds to the energy difference between the 
highest excited level and the ionization limit [15]. For this scheme, the energy difference 
between the 6 2 P° 3 / 2 level and the ionization limit is 2.44 eV which corresponds to 510 
nm. The argon ion laser can operate in single-line output mode and lase at 456 nm, 476 
nm, 488 nm, 496 nm, and 514 nm. Each of these wavelengths was investigated in this 
study. Figure 2-1 shows an energy level diagram of the two-step ionization scheme. 
Measurements were made of the ionization signal as a function of ionizing laser power; a 
linear relationship was observed. Calculations of the efficiency of this scheme are also 
discussed. 
Experimental Setup 

A schematic of the experimental setup is shown in Figure 2-2. The 6 2 Si/ 2 -> 
6 P°3/ 2 transition was laser pumped at 852.12 nm (ki) by an external cavity diode laser 
(Model 2010A, Newport Corp., USA) with a maximum output power of 16 mW. The 
beam was directed into the front window of a cesium ionization cell (Opthos Instruments, 
Rockville, MD, USA) made of pyrex. This cell contained a pair of planar, nickel 
electrodes near the front window separated by approximately 7 mm. The electrodes were 
connected to the exterior of the cell by a pair of electrical feedthroughs. The ionization 
cell was approximately 60 mm in length and contained 25 mm diameter windows. The 



13 



A> 

ionization limit 



496 nm 



6 2 P° 

° r 3/2 



852.12 nm 



6 2 S, 



Vi 



Figure 2.1. Partial energy level diagram of cesium showing the two-step ionization 
scheme. 



14 



Lock-in amplifier 



Oscilloscope 




Chopper 
Driver 



Power 
meter 



1 optical 



isolator 




700 nm 

longpass 

filter 



Cs cell 




Figure 2.2 Experimental setup showing two-step ionization in the cesium ionization cell. 



15 

cell was evacuated and contained a fill of solid cesium metal. The output of an argon ion 
laser (Model 2060-65, SpectraPhysics, Mountain View, CA) was directed through the 
rear of the ionization cell to overlap with X\. The argon ion laser was operated in single- 
line output mode and could be manually tuned to several wavelengths; each wavelength 
had a different maximum output power, ranging from 670 mW to 1.5 W. A 700 nm 
highpass filter and an optical isolator were used to protect the A-i diode laser from 
potential damage from the argon ion beam. A transimpedence amplifier (Model SR570, 
Stanford Research Systems, Sunnyvale, California) was used to apply a bias voltage to 
the electrodes and detect current flowing between the electrodes. An optical chopper was 
placed in the X\ beam path to modulate the ionization signal. This was necessary because 
a large nonselective ionization signal was observed that was due to the argon ion beam. 
The chopper was also used to trigger a lock-in amplifier (Model 128A, Princeton Applied 
Research, Oak Ridge, TN) that was used to demodulate the output of the current 
amplifier. The output of the lock-in amplifier was monitored with an oscilloscope. A 
microscope slide was placed in the argon ion beam path at a 45 degree angle to reflect a 
portion of the beam into an optical power meter (Ophir, Danvers, MA) to monitor the 
ionizing laser power. 
Results 

Four different argon ion wavelengths were evaluated for use as an ionization 
source. For each of the four wavelengths, a voltage reading was taken from the 
oscilloscope to measure the relative magnitude of ionization. Figure 2-3 shows a 
summary of these measurements. The ionization signal (mV) is shown as a function of 
energy difference (eV) between the lasing wavelength and the ideal wavelength at 510 



16 



" 








1.4- 










I 






1.2- 




I 




> 10 " 






I 


Signal 

o 
bo 
1 








a 
o 
-a 0.6 - 

TO 

N 

"3 

o 
•"- 1 0.4- 








0.2- 








0.0- 


1 i | i | r 1 1 1 


■ i i i i 


i ■ i ' i 



-0.050 -0.025 0.000 0.025 0.050 0.075 0.100 0.125 0.150 0.175 0.200 
AE = Photon Energy - Ionization Threshold (eV) 



Figure 2-3. Ionization signal obtained at four different lasing wavelengths. Each lasing 
output is scaled to 1 W output power for comparison. The 514 nm and 496 nm output are 
closest to the ideal ionization threshold. The error bars shown represent 7 replicate 
measurements. 



17 

nm. No ionization signal was observed for photons at 514 nm, while the largest ionization 
signal was measured using the 496 nm output (a difference of 0.06 eV). 

Since the 496 nm output of the argon ion laser achieved the highest degree of 
ionization, this wavelength was used for subsequent investigations using the two-step 
ionization scheme. Once this wavelength was chosen, the output power of the argon ion 
was incrementally attenuated to find the relationship between output power and 
ionization signal. Under ideal circumstances, the output of the ionizing laser would be 
large enough to ionize the maximum number of excited state atoms; under these 
conditions, the ionizing transition is optically saturated. To investigate such effects, the 
ionization signal using the 496 nm output was monitored as a function of the argon ion 
power reaching the cell. Figure 2-4 shows a linear response between ionization signal and 
ionizing power. If the transition were saturating, the response would plateau at the higher 
power levels. The argon ion beam was focused into the cell in order to increase the 
irradiance and attempt to saturate the transition. As a result, the ionization signal 
increased, but saturation of the transition was not observed. 

Saturation is difficult to achieve due to the low cross-sections for photoionization. 
For example, the photoionization cross-section for the ionizing transition using 496 nm 
photons is 1.8 x 10"' 7 cm 2 [16]. The argon ion laser operating at 496 nm had an output of 
1.9 x 10 photons/s/cm 2 . The product of these two quantities yields the pumping rate (34 
Hz). A transient, higher pumping rate could have been achieved by using a pulsed laser 
with a higher irradiance. 



18 



1400- 



1200- 



> 


1000- 


B 




13 




i 


800- 


C/3 




a 


• 


5 






600- 



400- 



200- 







I 



-1 — [— 

0.0 0.1 



I 



1 



i 



k 



0.2 



0.5 



0.3 0.4 

Power (W) at 496 nm 



0.6 



- 1 - 
0.7 



i 



0.8 



Figure 2-4. Ionization signal as a function of ionizing laser power at 496 nm. The error 
bars shown represent five replicate measurements. 



19 



To calculate the quantum efficiency of the two-step scheme, the ratio of the 
charges created to the number of photons absorbed should be determined. The ionization 
signal was measured using a relatively low power of X\ (0.054 mW). At this power, the 
number of \\ photons entering the cell is 2.3 x 10 14 photons/s. The fraction, a, of atoms 
absorbing these photons can be calculated as a, = \- Q~ a n ' where a is the 
absorption cross-section for the absorbing transition (2.4 x 10"" cm 2 ), n is the number 
density of ground-state atoms capable of absorbing the incoming photons (3 x 10 8 cm" 3 ), 
and / is the absorption pathlength under consideration (approximately 0.5 cm). The 
fraction of atoms absorbing under these circumstances is 0.003; from this value, the 
number of ~k\ photons absorbed is 6.9 x 10 11 photons/s. The corresponding ionization 
signal is obtained by the output of the lock-in amplifier (92.4 mV). When the 
amplification due to the lock-in and current-to-voltage amplifier is taken into account, the 
rate of charge collection at the electrodes is 1.3 x 10 8 electrons/s. The quantum 
efficiency, found by dividing these two values, is 2 x 10" 4 electrons/photon. 

Three-Step Ionization Scheme 

The first transition of the three-step ionization scheme is the same used in the 
two-step scheme (6 2 S,/ 2 -> 6 2 P 3 / 2 ). Atoms are then excited to the 6 2 D 5/2 state by laser 
excitation at 917.23 nm. The D-state is 1.084 eV from the ionization limit. A suitable 
laser to achieve photoionization from this state is a Nd:YAG operating at 1064 nm (1.167 
eV). Figure 2-5 shows an energy level diagram of this 3-step ionization scheme. 
Measurements were made of the ionization signal as a function of ionizing laser power. 
The presence of a large non-selective ionization signal was also observed in these studies; 



20 



ionization limit 




1064 nm 



6 2 D 



5/2 



917.23 nm 



6 2 P? 



852.12 nm 



6 2 S 



1/2 



Figure 2-5. Partial energy level diagram of cesium showing the three-step ionization 
scheme. 



21 

however, at lower laser powers the observed ionization signal was entirely selective. The 
calculation of the quantum efficiency of this scheme is also discussed. 
Experimental Setup 

The experimental setup used is shown in Figure 2-6. An external cavity diode 
laser (Model 2010A, Newport Corp., USA) at 852.12 nm (A-i) was directed into the front 
window of the cesium ionization cell. A second, external cavity diode laser (Model TEC- 
500, Sacher Lasertechnik, Marburg, Germany) at 917.23 nm (6 2 P 3 / 2 -> 6 2 D 5/2 ) was 
directed through the rear of the cell to overlap with X|. The ionizing laser was a pulsed 
Nd:YAG laser (Big Sky Lasers Technologies, Bozeman, MT). The output of the 
Nd:YAG was made to overlap with X { and X 2 inside the cell between the electrodes. A 
microscope slide placed at 45 degrees directed a portion of the beam to a photodiode. The 
photodiode signal was amplified and directed to an oscilloscope. The photodiode allowed 
for measuring the pulse-to-pulse reproducibility of the Nd:YAG. The response of the 
photodiode was calibrated using an Ophir power meter (Ophir, Danvers, MA) such that a 
1 mV photodiode signal on the scope represented a 0.092 mJ/pulse incident the ionization 
cell window. A bias voltage was applied to the electrodes using a transimpedence 
amplifier (Model SR570, Stanford Research Systems, Sunnyvale, California) which also 
detected the current passing between the two electrodes. The output of the amplifier was 
connected to the oscilloscope. The oscilloscope was triggered by the Nd:YAG power 
supply. 






22 



oscilloscope 



Nd:YAG 



V/A 






photodiode 
/■ 



transimpedance 
amplifier 



m 



=E 



_\ 



| optical isolator V_/S CCll optical isolator [~ 



X. 



X, 



Figure 2-6. Experimental setup showing three-step ionization in the cesium ionization 
cell. 



23 

Results 

The oscilloscope trace of the ionization signal at 5 mJ/pulse is shown in Figure 2- 
7. The larger trace shown represents the ionization signal when all three lasers are tuned 
and overlapping within the cell. The smaller trace represents the signal obtained when 
either X\ and/or X 2 are detuned from resonance (or blocked entirely). Ideally, when either 
of the diode lasers are detuned or blocked, the ionization signal would drop to the 
baseline. At 3 mJ/pulse, the relative contribution from the non-selective ionization signal 
decreased (Figure 2-8), and was not observed at a pulse energies less than or equal to 2.5 
mJ. Each data point shown in Figure 2-9 represents an individual oscilloscope trace. The 
ordinate of the plot represents the area of each ionization peak, which can be expressed as 
the number of electrons created during that pulse. At lower pulse energies, selective 
ionization is observed, while at higher energies (> 2.5 mJ/pulse) an increasing 
contribution from non-selective ionization is observed. The data points between 0.5 
mJ/pulse and 2.5 mJ/pulse are also plotted in Figure 2-10. The slight non-linearity 
between 2.0 mJ/pulse and 2.5 mJ/pulse indicates that the transition into the ionization 
continuum may be approaching saturation by the Nd:YAG while avoiding non-selective 
ionization. 

The Nd:YAG had an output of 4.7 x 10 24 photons/s/cm 2 (2.5 mJ/pulse ; 10 ns 
pulse width), which is several orders of magnitude higher than provided by the argon ion 
laser. A photoionization cross-section could not be found in the literature, although 
photoionizaton studies have been performed from the 6D levels at shorter wavelengths 
[17-19]. If the photoionization cross section is conservatively estimated as 10" 19 cm 2 , then 
a pumping rate of 5 x 10 5 Hz is achieved by the Nd:YAG laser. 



24 




20 

time ((is) 



r 
30 



Figure 2-7. Two oscilloscope traces of the ionization signal at 5 mJ/pulse. The larger 
trace is the ionization signal when all three lasers are tuned and overlapping within the 
cell. The smaller trace represents the signal obtained when X| is detuned from resonance. 
The ionization signal resembles the smaller trace when either A,, or X 2 is detuned or when 
the laser outputs are blocked. 




10 20 

time (|is) 



Figure 2-8. Two oscilloscope traces of the ionization signal at 3 mJ/pulse. The 
contribution the non-selective ionization signal (smaller peak) to the total signal 
decreases with decreasing pulse energy. 



26 


80- 




70- 


■ 




X 


"yT 


m 


> 60- 




o 




x 50- 




— 




? 40- 




D) 




C/) 




c 30- 




g 


■ 


"*j 




ca 




•S 20- 




c 




_o 




10- 






If 


0- 

( 





i ' i i ■ i 1 ' 1 

) 1 2 3 4 5 6 


Energy (mJ/Pulse) 


Figure 2-9. Ionization signal at various ionizing pulse energies. Each data point shown 


represents an individual oscilloscope trace. At lower pulse energies, selective ionization 


is observed, while at higher energies (> 2.5 mJ/pulse), an increasing contribution from 


non-selective ionization is observed. 



27 





7- 














'vt 


6- 






> 








l~- 


■ 






o 








•*- 


5- 






X 








■^ 






■ 


ro 








c 


4- 




_ ■ 


D) 






■ 


0) 








C 








g 


3- 






*-> 








03 








N 








'c 








o 


2- 










■ 






1- 


■ 






i i 


i ■ i 



0.5 



1.0 



t ■ 1 1 1— 

1.5 2.0 2.5 



Energy (mJ/Pulse) 



Figure 2-10. Ionization signal at lower pulse energies. 



28 

The quantum efficiency for this scheme is found by dividing the electrons created during 
a given pulse by the approximate number of X\ photons absorbed during the pulse 
duration. When 0.041 mW of X\ entered the cell, the number of photons during the pulse 
interval (10 ns) is on the order of 1.76 x 10 6 photons/s. The number of electrons can be 
found by integrating the ionization peak and considering the amplification factor of the 
amplifier; the number of electrons detected was approximately 2 x 10 4 electrons/pulse. 
The quantum efficiency based on these values is approximately 0.01 electrons/photon. 

Fluorescence Detection Scheme 

The first absorption step in fluorescence detection is the same transition used in 
the ionization schemes discussed above, at 852.12 nm (6 2 Si/ 2 -> 6 2 P° 3/2 ). This is followed 
by laser-pumped excitation at 917.23 nm (6 2 P 3 / 2 ■> 6 2 D 5/2 ) followed by radiative decay 
to the 7 2 P°3/ 2 level. Atoms in this level then fluoresce (7 2 P° 3/2 -> 6 2 S ]/2 ) to the ground 
state, emitting photons at 455.53 nm. In resonance fluorescence detection, absorption of 
photons by the first transition (852.12 nm) is monitored by detecting the fluorescence at 
455.53 nm. The atomic levels involved in this process are shown in Figure 2-11. 

The sensitivity of spectrally selective photon detection using an RFD or RFIM is 
limited by the efficiency of converting 852.12 nm photons into photons at 455.53. In 
order to evaluate this figure of merit, a photomultiplier tube (PMT) was calibrated to 
quantify the relationship between the PMT's current output and the number of photons 
incident on the photocathode. Once calibrated, this PMT was then used to make 
fluorescence measurements at 455.53 nm, when both 852.12 nm and 917.23 nm lasers 
were directed through a cesium cell. The quantum efficiency of this scheme was then 
found by comparing the number of photons corresponding to the input (852. 12 nm) and 



29 




6 2 P° 

° r 3/2 



6 2 D 



5/2 



Figure 2-11. Partial energy level diagram of cesium showing two-step excitation scheme 
for resonance fluorescence detection. 



30 

output (455.53 nra) wavelengths. 
Calibration of Photomultiplier Tube 

The PMT (Model 1547, Hamamatsu, Japan) used in this study was calibrated 
using a NIST-traceable 1000 W tungsten lamp source (Oriel Instruments, Stratford, CT). 
The lamp was certified at a various wavelengths; at 455 nm, the tungsten lamp had a 
spectral irradiance of 4.62 mW/cm 2 /nm at a distance of 50 cm. A large piece of sheet 
metal with a small, circular aperture (Area = 0.00785 cm 2 ) was placed 50 cm away from 
the tungsten lamp. Through such arrangement, the amount of light passing through the 
pinhole could be determined. The PMT and associated optics were placed at the opposite 
side of the metal sheet. One consideration in this calibration was the alignment of the 
aperture, pinhole, and PMT in the same orientation used in the detection of 455.53 nm 
fluorescence from a cesium cell. To this end, the aperture was considered the source of 
light, and a lens was placed to generate a 1:1 image of this aperture on the surface of the 
PMT. A 2.0 neutral density filter was placed in front of the PMT to further decrease the 
number of photons incident the PMT. A 455 nm interference filter (56% transmission at 
455 nm; 9.94 nm FWHM) was placed in front of the PMT housing. If the spectral 
bandwidth of interest was defined by the FWHM of the interference filter, the power of 
the light passing through the aperture within this spectral bandwidth was 0.36 uW. The 
power incident on the PMT was calculated by accounting for the following losses: losses 
at front and back surface of the lens, transmission of the neutral density filter, and 
transmission through the interference filter. Once these losses were accounted for, the 
power incident on the PMT was 9.30 x 10 -4 uW. The current generated by the PMT was 
measured with an electrometer. The background-subtracted signal created by 9.30 x 10" 4 



31 

uW of incident light was 59.5 p.A. Dividing these two values resulted in a calibration 
value of 6.40 x 10 AAV. The calibration value was compared to values obtained from the 
manufacturer's catalog. The product of the cathodic sensitivity (45 mA/W) and the gain 
(3 x 10 ) is 1 x 10 5 AAV, which is less than a factor of 2 different from the experimentally 
determined value. This latter value may not be representative of the calibration performed 
since the cathodic sensitivity was not listed at a specific wavelength; in addition, the 
value for the gain was given for 1000 V applied to the tube. In these experiments, the 
power supply used provided an output of 935 V. 
Measurement of Resonance Fluorescence 

Once the PMT was calibrated, a cesium cell was placed in the position of the 
aperture described above in order to measure the 455.53 run intensity in absolute units. 
An illustration of this experimental setup is shown in Figure 2-12. The lens, neutral 
density filter, and interference filter remained in the same position since they were 
included in the PMT calibration. An external cavity diode laser (A-i = 852.12 nm) (Model 
20 10 A, Newport Corp., USA), with an output power of 16 mW and a linewidth of 5 
MHz, was tuned to the cesium 6 2 Si /2 -> 6 2 P 3 / 2 transition. The output of this laser was 
directed through the front window of the cell. A second external cavity diode laser (k 2 = 
917.23 nm) (Model TEC-500, Sacher Lasertechnik, Marburg, Germany) was used to 
pump the 6 ?° V2 -> 6 D 5/2 transition. The cesium cell (Opthos Instruments, Rockville, 
MD) contained a fill of solid cesium under vacuum; the cell was 75 mm in length by 25 
mm in diameter. The X\ and X 2 beams were overlapped within the cesium cell such that 
beam overlap occurred in the same position as the aperture used in the PMT calibration. 
Optical isolators were placed at the exits of both diode lasers to protect the lasers from 



32 



Laser 



Optical Isolator 



Interference Filter 

I 



pPMT 

Lens t 

Neutral 

Density Filter 
Optical Isolator 



Electrometer 



K 

Laser 



Figure 2-12. Experimental setup showing two-step excitation and measurement of 455.53 
nm fluorescence. 



33 

reflections entering the cavity. 
Results 

The PMT current measured when ^i and X 2 were tuned to resonance was 0.0476 
uA. Using the conversion factor above (6.40 x 10 4 AAV), a power of 7.44 x 10" 7 p.W was 
incident the photomultiplier tube. This power at 455.53 nm corresponded to 1.68 x 10 6 
photons/second. When corrected for the limited collection of light and other losses (lens 
and filters), the number of photons emitted per second from the cell was on the order of 5 

Q 

x 10 photons at 455.53 nm. To determine the quantum efficiency, the number of 
fluorescent photons should be divided by the number of 852.12 nm photons absorbed 
within the cell (6.3 x 10 11 photons/second). The quantum efficiency based on these 
values was on the order of 8 x 10" 4 . 

Conclusions 

The efficiencies of three schemes for use in cesium based photon detectors have 
been evaluated. The two-step scheme based on photoionization at 496 nm has a quantum 
efficiency of 2 x 10" 4 electrons/photon. The three-step scheme using a Nd:YAG at 1064 
nm has a larger quantum efficiency (0.01 electrons/photon) due to the large irradiance 
provided by this pulsed source, but this scheme suffers from a diminished duty cycle. 
However, efficient photoionization may likely require a pulsed laser to pump the 
photoionizing transition. In addition, photon detection using such a pulsed laser would 
benefit from pulsed excitation lasers so that the absorption of Xi photons temporally 
coincides with further excitation and ionization steps. 

The quantum efficiency using the resonance fluorescence detection scheme was 8 
x 10 . This efficiency is much lower than that theoretically achievable by successful 



34 

implementation of ionization detection [20]. Although theoretically superior, the practical 
and successful implementation of cesium ionization imaging remains a task facing 
several engineering challenges. In light of these difficulties, the remainder of this work 
will be described in the context of the resonance fluorescence detection scheme described 
above. Fortunately, the improvements in spectral resolution and applications discussed in 
later chapters can be readily applied to either ionization or fluorescence detection. To 
further understand the RFD process, resonance fluorescence detection will be evaluated 
theoretically in the following chapter to identify processes that compete with the desired 
relaxation pathway. 






CHAPTER 3 
THEORETICAL EVALUATION OF A CESIUM RESONANCE FLUORESCENCE 

DETECTION SCHEME 

Introduction 

The steady-state and temporal behavior of laser interaction with an atomic system 
are often described by either the rate equation formalism or the density matrix formalism 
[21-26]. The density matrix formalism provides the more complete description of laser- 
excited atomic systems for several reasons including its proper treatment of coherence 
effects due to the finite bandwidth of the source interacting with the absorption profile. 
While the density matrix formalism is the more accurate of the two methods, it is also 
computationally and conceptually difficult compared to the rate equation approach. The 
rate equation formalism is simpler, although this method disregards several properties of 
the system, namely the source and absorption profiles, coherent effects, and temporal 
behavior. The rate equation approach also neglects certain effects caused by narrowband 
light at high intensities, such as dynamic Stark effects (broadening, splitting, and shifting 
of levels) and two-photon excitation [14,26]. In general, these effects are significant 
whenever high intensity, narrowband light excites atoms in a weakly collisional media 
[26]. The rate equation approach should give a reasonable description of an atomic 
system whenever narrowband light of relatively low intensities is used since the effects 
described above should be minimized. 

We discuss the use of rate equations to solve an eight-level system under steady- 
state conditions. The eight-level system is based on the two-step resonance fluorescence 

35 



36 

monochromator (RFM) scheme experimentally evaluated in Chapter 2. Although only 
four levels are necessary for the RFM scheme described, the remaining four levels 
contribute to possible relaxation pathways that could compete with resonance 
fluorescence detection. The density matrix formalism (using a reduced-level system) is 
also used in this study. In either the rate-equation or density matrix simulations, the 
fraction of atoms populating each level is determined. Through these population 
fractions, competing fluorescence pathways can be determined. Additionally, the 
quantum efficiency can be determined, and this value is compared to experimental 
values. 

Methodology for Rate Equation Model 
Cesium Excitation Scheme 

The cesium system under investigation was modeled using MathCad 8 
Professional (MathSoft, Inc.,). A partial energy level diagram is shown in Figure 3-1. In 
the cesium excitation scheme shown, ground-state cesium atoms (6 2 Si/2) are optically 
pumped to the 6 2 P°3/ 2 level by an external-cavity diode laser with a linewidth of ~ 5 MHz 
operating at 852.12 nra. Further excitation to the 6 2 Ds/2 state is made with a similar diode 
laser operating at 917.23 nm. A variety of relaxation pathways are possible from the 
6 D 5/2 state. The pathway of particular interest is the 7 2 P°3/2 H> 6 2 Si/2 radiative transition 
at 455.53 nm. This fluorescence is the detection wavelength in the cesium RFM. Briefly, 
the RFM detects photons lying within the narrow absorption band of the first transition 
(6 S1/2 -> 6 P°3/2)- The excited atoms are pumped to the 6 2 D 5 / 2 state through laser 
excitation at 917.23 nm. This process is detected through measuring the fluorescence at 
455.53 nm, which is the final step in the 6 2 D 5/2 -> 7 2 P° 3 /2 -> 6 2 Si/2 pathway to the ground 
state. The other optical pathways to the ground state, shown in Figure 3-1, compete with 











37 






' 1 V^rP 


<apr 6 2 d 5/2 


^r ' ^1/2 / 


^/ N 


\ 




O)/ 


J^U 5 2 D 5 A 
if 52 ° 32 ) 










Vr^-6 : P" # 

Mi m /f 




§/ / 






ai /j 






%l £// 






<o'# ^// 






£>1 ^1/ 






V / *v#/ 












V/i£l&£: 6 c 






" r "^ D °l/2 


Figure 3-1. Partial energy level 


diagram of cesium 


showing the excitation/fluorescence 


wavelengths along with other possible radiative pathways to the ground state. 



38 

the RFM scheme. The quadrapole transitions, n 2 Ds/2 -^ 6 2 Si/2 and n 2 D3/2 -^ 6 2 Si/2, are not 
expected to significantly contribute to the population distribution among the several 
levels; however, they were included for completeness. The coefficients for spontaneous 
emission for all transitions involved in this cesium scheme are given in Table 3-1 [27]. 
These coefficients are the basis for the rate equation model described below. 
Rate Equation Model of Eight Atomic Levels 

A series of rate equations were generated using the various transitions of the 
atomic system. Each of the eight atomic levels in the cesium scheme represents one of the 
rate equations shown in the Appendix. For any given level under steady-state conditions, 
the difference between the processes contributing to the population of a level and the 
processes depleting the population is assumed to be zero. These processes include the 
rates of spontaneous emission and the pumping rates for the two laser-connected 
transitions. 

In addition to the eight equations, a ninth equation is also used in this model. This 
equation states that the sum of the populations of the eight levels should equal the total 
population under study. This assumption is reasonable because losses within the system, 
such as ionization losses, are minimal. In addition, at a constant temperature, the number 
density and distribution of the Cs vapor remains constant. 

The nine equations were entered into the MathCad (MathSoft, Inc., U.S.A.) 
Minerr function and were solved to find the population of each atomic level. This 
function calls upon an iterative algorithm, called the Levenberg-Marquardt method [28], 
in order to generate solutions that satisfy all the equations. When solving a system of 
equations numerically, MathCad requires the user to define initial values from which to 
start the iteration process; it also allows the user to provide initial constraints on the 



39 



Table 3-1. Coefficients (A) of spontaneous emission (s" 1 ) for transitions studied [27]. 



Atomic Transition 


A (s" 1 ) 


6 P°3/2 "> 6 S1/2 


3.70x10' 


6^5/2 -> 6 2 P° 3 /2 


1.74 x 10' 


6 2 D 5/2 -> 7 2 P°3 /2 


6.67 x 10 4 


6 D 5/2 ■> 6 S1/2 


2.01 x 10' 


7 2 P° 3/2 ■> 6 2 S 1/2 


4.186xl0 6 


7T M -» 7 2 S 1/2 


3.98 x 10 6 


7 2 P° 3/2 -> 5 2 D 5/2 


8.35 x 10- 


7 P 3/2 "^ 5 D3/2 


7.52 x lO 4 " 


7 2 S 1/2 "* 6 2 P° 3 /2 


1.22 xlO 7 


7 2 S 1/2 -> 6 2 P° 1/2 


5.53 xlO 6 


5 D5/2 -> 6 S1/2 


2.07x10' 


5 2 D 5/2 -> 6 2 P 3/2 


8.4 x 10 5 


5 2 D 3/2 -» 6 2 P° 3/2 


1.22 x 10 5 


5 2 D 3/2 -» 6 2 P° 1/2 


1.0 xlO 6 


5 D3/2 ■> 6 S1/2 


1.96 xlO -1 


6 P°i/2 "> 6 S1/2 


2.89 x 10 v 



40 

solution set. Typically, the initial values were set such that the majority of the population 
was divided among the three laser connected levels, while the remaining population was 
evenly distributed in the remaining levels. Typical constraints on the solution set included 
some basic assumptions, such as limiting the solution set to contain only positive values. 
When reasonable values and constraints were set, the Minerr function generated a 
solution set within ten seconds on a personal computer. 

The solution set was then stored as a vector; the members of this array were then 
checked for accuracy and appropriateness. Such confirmation is necessary because the 
Minerr function returns the solutions that are closest to satisfying the set of equations 
whenever it cannot find an exact solution set. In order to test the solutions, the elements 
of the solution set vector were then substituted into the original set of equations. The sum 
of the positive terms was compared to the sum of the negative terms for each equation. 
For an exact solution, the magnitude of these two sums would be equal; however, in the 
worst cases, they differed by only 0.5% for a given equation. 
Results 

The laser output exciting the 6 2 Si/ 2 -> 6 2 P°3/2 (k\) transition could be varied by 
changing the corresponding pumping rate throughout the eight equations. Keeping with 
the context of resonance fluorescence detection, a relatively low power of X\ (0.05 mW) 
was used in the worksheet shown in the Appendix. The fraction of the total population 
attributed to each of the eight levels at this laser power is shown in Table 3-2. Emission 
intensities (photons/s) for transitions originating from these levels can be found by 
multiplying the Einstein coefficients for spontaneous emission (s" 1 ) and the number of 
atoms in the emitting level. The number of atoms in each level can be calculated from the 



41 



Table 3-2. Fraction of atoms excited to the given levels at steady-state conditions. 



Atomic level 


Fraction of total population 


6 Si/2 


0.998 


6 2 P°3/2 


1.5 x 10" J 


6'D 5/2 


2.0 x 10" 4 


7 2 P°3/2 


1.5 xlO -6 


l%a 


3.4 x 10"' 


5 2 D 5/2 


4.3 x 10"' 


5 2 D 3/2 


1.0 xlO" 7 


6 2 P°,/ 2 


6.8 x 10" 8 



42 

fractions shown in Table 3-2 if the total number of atoms is known. This value can be 
approximated by considering the number density of cesium at room temperature (3x10 
atoms/cm ), the fraction of atoms having profiles matching the narrowband laser source 
(~ 0.01), and the interaction region of the two-step excitation (~ 0.5 cm 3 ). Once the 
population fractions have been converted (approximately) into the number of atoms, the 
emission intensity for each transition can be found. Table 3-3 shows the relevant atomic 
transitions along with the emission intensity (in photons/s and nW). The transition of 
particular interest in the cesium RFM is the 7 2 P°3/2 "^ 6 2 Si/2 transition which emits 
approximately 0.5 nW (1.3 x 10 9 photons / s) under conditions described above. The ratio 
between this power and the 852.12 nra incident power, 0.05 mW, is 10" 5 . When the ratio 
is calculated based on the photons emitted and absorbed per second, the ratio becomes 2 
x 10" , which is reasonable when considering the experimentally determined quantum 
efficiency of 8 x 10" 4 reported in the previous chapter. 

The 6 D 5 / 2 -> 7 2 P°3/2 transition is the sole process that populates the 7 2 P 3 / 2 level; 
however, this transition radiatively emits only a small power since the photons emitted at 
this transition have such a large wavelength (14.35 |im). Once the 7 2 P°3 /2 level is 
populated, four possible radiative pathways are considered in the model. In the context of 
resonance fluorescence detection, one of these pathways is of considerable importance 
(7 P°3/ 2 "^ 6 S1/2 at 455.53 nm), while the other three pathways are only of concern if 



they significantly compete with the RPM process. In terms of power output, the 7 2 P 



2 n O 



3/2 



7 S1/2 transition emits at approximately 20% the power of the 7 2 P°3/ 2 "> 6 2 Si/ 2 transition. 
It is worth noting that nearly an identical number of photons are emitted per second by 
each of these two transitions; the discrepency in the output power is due to the large 



43 



Table 3-3. Power emitted from each transition shown. 



Atomic Transition 


Wavelength 
(run) 


Power (nW) 


6'Dsn ■> 7 2 P° 3 /2 


14350 


0.04 


7 2 P° 3 /2 ■» 6 2 S 1/2 


455.53 


0.5 


7 2 P°3/ 2 -* 7 2 S 1/2 


2930.8 


0.08 


7 2 P° 3 /2 -> 5 2 D 5/2 


1360.2 


0.04 


7 2 P°3/ 2 * 5 2 D 3/2 


1342.4 


0.003 


7 2 S 1/2 -> 6 2 P°3/ 2 


1469.5 


0.1 


7 2 S 1/2 -> 6 2 P° l/2 


1358.8 


0.05 


5 D5/2 -> 6 S1/2 


684.9 


5.15 x 10" 7 


5 2 D 5/2 "» 6 2 P°3/2 


3490.1 


0.004 


5 2 D 3/2 "» 6 2 P°3/2 


3616.1 


0.0001 


5 2 D 3/2 -» 6 2 P° l/2 


3010.3 


0.001 


5 D3/2 ■> 6 S1/2 


689.5 


1.1 x 10" 9 


6 P°i/2 -> 6 S1/2 


894.35 


0.09 



44 

difference in wavelengths between these two tansitions (2930.8 nm compared to 455.53 
nm). The other two relaxation pathways do not compete as significantly with the RFM 
process. 

Any of the transitions shown in Table 3-3 could be used as the fluorescence 
detection wavelength to monitor 852.12 nm absorption in the RFM; however, the 
emission at 455.53 nm is the most intense and occurs at a convenient wavelength for 
detection by photomultiplier tubes (which readily provide high quantum efficiencies for 
detecting light in this wavelength region) and camera systems. 

Density Matrix Simulations 

In order to evaluate the predictions made by the rate equation model described 
above, the density matrix formalism was used to model a reduced system. Density matrix 
simulations were performed using the software package, Densmat [29], to model a 
reduced system consisting of the three laser coupled atomic levels (6 2 Si/2 -> 6 2 P° 3 / 2 -> 
6 D 5/2 ). The same parameters used in the rate equation model were used in the density 
matrix simulations (wavelengths, laser powers, linewidths, and Einstein coefficients). 
Since the density matrix software used is configured for pulsed excitation, the cw lasers 
used in these studies were treated as a long pulse of 25 us. This approximation is 
reasonable if steady-state conditions are achieved relatively quickly with respect to the 
duration of the fictitious laser pulse [30]. The duration of the simulations shown in Figure 
3-2 and Figure 3-3 were 800 ns, and the three levels shown reached steady-state 
conditions within 300 ns. Figure 3-2 shows the temporal behavior of the 6 2 Si/ 2 level until 
a steady-state population fraction of 0.999 is achieved. The results for the 6 2 P 3 / 2 and 



45 



1.0000 



g °" 98H 

I 

3 

o 0.9996 

O 

.8 

S 0.9994 



0.9992- 



0.9990 







200 



1 1 

400 

Time (ns) 



600 



800 



Figure 3-2. Density matrix simulation showing depletion of 6 2 Si /2 level until steady-state 
conditions are reached. 



46 



8 
7 
6H 



a 

I « 

I, 

Oh 
O 
§ 



2- 



1 1 



0- 



-1 



o 



200 



1 

400 

Time (ns) 



6 2 P° 




600 



800 



Figure 3-3. Density matrix simulation showing the 6 2 P 3 / 2 and 6 2 D 5/2 levels reaching 
steady-state conditions. 



47 



6 D 5/2 levels are shown in Figure 3-3; these two levels reach a steady-state population 
fraction of 7.7 x 10" 4 and 1.2 x 10" 4 , respectively. These values are reasonably close the 
rate equation values (15 x 10" 4 and 2.0 x 10" 4 , respectively). 

Conclusions 
A model based on steady-state rate equations has been described for an eight-level 
cesium system. The power emitted by several transitions was described; the 6 2 D 5/2 
->7 P°3/ 2 transition at 455.53 nm had the highest emission, while other pathways had 
smaller roles. The quantum efficiency determined from the rate-equation model was in 
reasonable agreement with experimental results. In addition, population fractions 
determined by the reduced-level density matrix model were in reasonable agreement with 
the rate-equation model. It is interesting to note that the second and third strongest 
transitions (7 2 P° 3 /2 ■> 7 2 Si/ 2 and 7 2 S, /2 -> 6 2 P 3 / 2 ) place atoms in the 6 2 P° 3/2 state. Some of 
the atoms decaying by this pathway will not be lost, since upon relaxation to the 6 2 P° 3/2 
state, these atoms can be excited by X 2 (917.23 nm) into the system once again. 
Alternatively, the atoms could relax into the ground state (6 2 P° 3/2 -> 6 2 Si/2) emitting 
photons at 852.12 nm; these photons could either be emitted from the system or be 
reabsorbed. 

The quantum efficiency predicted by the rate-equation model differed from 
experimental values by a factor of 2.5. The rate-equation model described is not capable 
of providing a complete description of the processes occurring when cesium atoms are 
excited by narrowband laser sources. Even though the rate-equation model does not 
provide an entirely accurate account, the competing relaxation pathways can be labeled 



48 

as either "significant" or "insignificant" by using such a model; such labeling is 
reasonable if the predictions are accurate within an order of magnitiude. 

The quantum efficiency of the cesium resonance fluorescence detector has been 
evaluated experimentally and theoretically. A second figure of merit, the spectral 
resolution, is also dependent upon the spectroscopy within the atomic vapor. Atomic 
vapor image detectors reported in the literature have operated with a spectral resolution 
limited by the Doppler-broadened profile of the atomic vapor [5-14]. In the next chapter, 
cesium-based non-imaging and imaging photon detectors are reported that operate with a 
sub-Doppler spectral resolution [3 1-32]. 



CHAPTER 4 

SUB-DOPPLER SPECTRAL RESOLUTION FOR A CESIUM RESONANCE 

FLUORESCENCE DETECTOR AND IMAGING MONOCHROMATOR 

Introduction to Sub-Doppler Excitation 

The velocity-selective interaction of a narrowband-laser with a Doppler- 
broadened atomic transition is investigated by saturation spectroscopy [33-40]. When a 
narrowband laser source interacts with a Doppler-broadened transition, only a selected 
portion of atoms within the laser profile will absorb the narrowband radiation. If the laser 
source is tunable, it can be used to excite a limited subset of this velocity distribution. If a 
second laser is coupled to one of the levels involved in this selective excitation, this 
second absorption process can be used to probe the selected population excited by the 
first laser. For example, if the first laser is tuned to the center of a Doppler-broadened 
transition, the second laser can be frequency scanned across a second transition (coupled 
to either the lower or higher atomic level); the resulting absorption profile (as the second 
laser as scanned) will contain either a narrow hole or a peak representing an observable 
excess or depletion in the atomic population that is due to the excitation by the first laser. 
This idea can be extended to the case when a multi-step atomic excitation is desired. If 
the first and second transitions are excited by narrowband lasers, the lasers can be tuned 
to excite the same velocity distribution of atoms. This is readily accomplished by 
arrangement of the beams in either co- or counter-propagating geometry. When both 
lasers are arranged in this geometry, and tuned to the centers of the Doppler-broadened 



49 



50 

profile, only the atoms with a zero net-velocity with respect to the beams will be excited 
to the highest state. 

We report on a two-step excitation scheme for a cesium resonance fluorescence 
monochromator (RFM) in which a sub-Doppler photon detection is achieved through 
beam alignment in co- and counter-propagating geometries. The RFM scheme has a 
detection wavelength at 852.12 nm (6 2 Si /2 -> 6 2 ? 3a ) followed by further excitation at 
917.23 nm (6 P%2 "> 6 2 D 5 / 2 ). The absorption process can be monitored by the 
fluorescence at 455.53 nm (6 2 P 3 / 2 -> 6 2 S] /2 ). The improved spectral resolution is 
accompanied by frequency tuning capabilities accomplished by tuning the X 2 (917 nm) 
laser to excite atoms with varying velocity components. The improved spectral resolution 
(200 MHz) and tuning capabilities for selective photon detection are demonstrated using 
a single-point, resonance fluorescence monochromator (RFM) [31]. 

The narrow spectral resolution achieved for the cesium RFM has also been 
applied toward image detection [32]. A cesium based RFIM has been previously 
developed in our laboratory [12] with a spatial resolution of 150 urn and a spectral 
resolution limited by the Doppler broadened absorption profile of cesium (400 MHz at 
26°C). Similar to the RFM described above, the RFIM detects (image-carrying) photons 
at 852.12 nm. The excitation process is the same as the single-point, RFM, except the 
excitation laser (917 nm) is expanded into two dimensions. As a result of the two-step 
excitation, the 455.53 nm fluorescence is emitted and can be detected as two dimensional 
image. Images of this blue fluorescence were captured using standard optics, an image 
intensifier, and digital CCD camera. The two-dimensional image plane within the cesium 
vapor was defined by the X 2 (917 nm) laser whose output was transformed into a sheet of 



51 

light and directed into the cesium cell orthogonal to the X\ (852 nm) input beam. In this 
excitation scheme and geometry, the two-step excitation resulted in a spectral response 
limited by Doppler broadening. Alignment of the two lasers used in the excitation 
scheme in a co-propagating geometry resulted in a detector spectral bandpass (270 MHz) 
that is narrower than the Doppler broadened profile (400 MHz) of cesium vapor at room 
temperature. Images of the cesium D 2 line emission (852.12 nm) from a cesium hollow 
cathode lamp were compared to an argon emission line (4S°i /2 -> 4P 3/2 ) at 852.14 nm 
from a silver/argon hollow cathode lamp as a demonstration of the spectral resolution. 
The ability of the RFTM to discriminate between photons exciting the F = 4 versus F = 3 
ground state hyperfine level (a frequency difference of 9.2 GHz) of the cesium 6 2 Si/ 2 
state was also investigated to further demonstrate the spectral selectivity. 
Photon Detection With Sub-Doppler Spectral Resolution 
A single-point, cesium resonance fluorescence detector was evaluated in three 
beam geometries: orthogonal, co-propagating, and counter-propagating. Fluorescence 
measurements at the detection wavelength (455.53 nm) were made as a function of A-i 
(852 nm) laser wavelength. These excitation profiles represented the spectral instrument 
function of the RFM. Orthogonal geometry was evaluated to confirm the previously 
described (Doppler-broadened) spectral resolution [12]. Co- and counter-propagating 
geometries were used to determine the narrowing effects due to velocity-selective 
excitation. In these geometries, k 2 (917.23 nm) was manually tuned to fixed values for a 
given A,i scan. A sub-Doppler spectral resolution that was tunable within the Doppler- 
broadened profile was observed. 



52 

Experimental Setup 

The experimental setup used to investigate the various beam geometries is shown 
in Figure 4-1. A tunable external cavity diode (X\) laser (Model 2010A, Newport Corp., 
USA), with an output power of 16 mW and a linewidth of 5 MHz, was tuned over the 
cesium 6 S1/2 ■> 6 P 3 / 2 transition by deflecting a piezoelectric element placed behind the 
external cavity tuning mirror. The piezoelectric was driven by a 1 Hz triangle wave using 
a function generator (Model F3GB, Wavetek, USA). A second tunable external cavity 
diode (k 2 ) laser (Model TEC-500, Sacher Lasertechnik, Marburg, Germany) was used as 
the pump laser at 917.23 nm (6 2 P 3 / 2 -> 6 2 D 5/2 ) and had a power of 30 mW and a 
linewidth of 5 MHz. The X\ and X 2 beams were each passed through optical isolators and 
then directed into a cesium vacuum cell (Opthos Instruments, Rockville, MD, USA) to 
spatially overlap in either orthogonal, co-propagating, or counter-propagating geometries. 
The cell was 75 mm in length by 25 mm in diameter. The intensity of the 852.12 (6 2 P 3 / 2 
-> 6 S1/2) or 455.53 (7 P° 3/2 -> 6 2 Si/ 2 ) fluorescence was collected by a lens, and directed 
through a 852 nm or 455 nm interference filter coupled to a photomultiplier tube (PMT) 
(R928, Hamamatsu, Japan). The PMT was coupled to a current to voltage amplifier and 
oscilloscope (Model TDS 3012, Tektronix, USA). 



53 




\RM2 



Figure 4-1. Experimental setup to investigate the excitation profile in three beam 
geometries. FG = function generator; 01 = optical isolator; Ml, M2, M3 = mirrors; RM1, 
RM2 = removable mirrors; BS = beam splitter; IF = 852 nm or 455 nm interference filter; 
PMT = photomultiplier tube; V/A = amplifier; OSC = oscilloscope. The A,, beam enters 
the cell from the left (Ml and BS). In co-propagating geometry, RM1 is removed and X 2 
is directed into the cell by the beam splitter (k 2 path = RM2, M2, BS). In counter- 
propagating geometry, RM2 is removed and X 2 is directed into the cell by M3. In 
orthogonal geometry, RM2 is removed, and X 2 is directed into the cell by RM1. 



54 

Results 

When the beams were aligned in orthogonal geometry, the full width at half 
maximum (FWHM) of the 852 nm and 455.53 nm excitation profiles were limited by 
Doppler broadening (400 MHz). The 455.53 nm fluorescence intensity as a function of Vi 
frequency in orthogonal geometry is shown in Figure 4-2. The 852 nm fluorescence was 
Doppler broadened for the co- and counter-propagating geometries as well; however, 
sub-Doppler profiles of the 455.53 nm fluorescence were observed for the co- and 
counter-propagating geometries. Figure 4-3 shows the 455.53 nm fluorescence detected 
for three Vi frequency sweeps when the beams were co-propagating. The Vi frequency 
scan was the same for all three waveforms; however, each waveform differed by the v 2 
frequency that was manually tuned to fixed values. The v 2 frequency was tuned by fixed 
increments; however, calibration data relating the frequency of v 2 at each increment were 
not available. The FWHM of the three excitation profiles shown are on the order of 250 
MHz and were tunable over the cesium 6 2 Si/ 2 -> 6 2 P° 3 /2 absorption profile. Since the 
spacing between the five D-state hyperfine levels involved is on the order of 25 MHz, the 
D hyperfine levels remain unresolved; these closely spaced hyperfine levels are expected 
to contribute to the peak widths. 

Several frequency traces are shown in Figure 4-4, with the beams aligned in 
counter-propagating geometry. The FWHM of each trace was on the order of 200 MHz. 
As v 2 was manually tuned to fixed frequencies, a variety of excitation profiles were 
observed as a function of Vi frequency sweep. The v 2 increments in this geometry were 
half the magnitude of those used in co-propagating geometry. Each trace is due to 
excitation from a specific P hyperfine level with contributions from adjacent hyperfine 



55 




0.6 


A 




*3 0.5 


\ 




3> 


/ \ 




"35 0.4 

c 

CD 


/ \ 




Fluorescence 

o o 

i , i 


J \ 




0.1 


^/ V 




-i 






w.u -\ . | ■ | i | i | 

-1.0 -0.5 0.0 0.5 1.0 


1 1 ' 1 

1.5 2.0 


Frequency (GHz) 




Figure 4-2. Fluorescence intensity (455.53 nm) as a function 
orthogonal beam geometry. 


of Vi frequency in 











56 




1.0- 






0.8- 






^ 






d 






TO 






r o.6- 






-i— > 






CO 






c 






Q) 






-t-> 






c 






CD 0.4- 


/ \ •' '• 




o 


/ V • » 




c 


/ \ •' 







/ \ ' 




o 


/\ / \ ■' 




w 


/ / A '• 




CD 

O 0.2- 

13 




U_ 


is \ \ 






Ijtr N s _ > ^ V. 




0.0- 






i • i i i 

-0.3 0.0 0.3 0.6 


1 
0.9 


Frequency (GHz) 




Figure 4-3. Fluorescence intensity as a function of Vi frequency in co 


-propagating beam 


geometry. Each trace represents a fixed increment of v 2 frequency. 





57 





1.6-1 




1.4- 




■ 




1.2- 


^^ 


. 


3 




m 


1.0- 






fc 


■ 






CO 

C 


0.8- 


C> 








£ 




W 


0.6- 


Q 




9 




B 




o 


0.4- 


1) 




C 




5 


• 


3 




PL, 


0.2- 




0.0- 




-0 2- 



■^^■ w a gWH i w i^pw ^B j jM «i> f nwwu w w iwl x i 



— *^mwj>^M* m» i> w *<^ w<*wi m^M&m 



w Nw i wmNwn i » «i n ii*»»ii n i m n nji 




MMMM«PM*MAMMMM«MMtfM 



.. ■■ |l ■«, ■■■ . — .,.. ,, — M ^, |,„ _ I,,- . 



~~ 1 ' 1 - 

-4 -2 

Frequency (GHz) 



Figure 4-4. Several V| frequency sweeps in the counter-propagating beam geometry. 
Each plot represents a fixed increment of V2 frequency. 



58 

levels that have been velocity-shifted into resonance. The use of co- and counter- 
propagating beam geometries has allowed for tunable photon detection with a spectral 
resolution as low as 200 MHz using an atomic vapor detector. The use of various beam 
geometries could serve as an enhancement to existing orthogonal-geometry RFIMs. 
Imaging with Sub-Doppler Spectral Resolution 

The improvements in spectral resolution in the RFM were also applied to the 
cesium RFIM. In order to accomplish the enhanced spectral resolution in the imaging 
detector, the excitation laser beams were arranged in a co-propagating geometry as 
opposed to the orthogonal geometry used previously [12]. In order to demonstrate the 
spectral background rejection capabilities of the RFIM, images acquired from the cesium 
D 2 emission from a cesium hollow cathode lamp were compared to images obtained from 
an intense argon emission line located 12 GHz away from the center of the cesium D 2 
line. We also report on the ability to discriminate between photons exciting the two 
hyperfine levels (separated by 9.2 GHz) of the cesium ground state. Images were also 
acquired of the X| (852 run) laser tuned to various positions along the Doppler-broadened 
profile. Such tuning allowed for a direct comparison of the spectral resolution in 
orthogonal and co-propagating geometries. 
Experimental Setup 

Two tunable, continuous-wave diode lasers in Littman-Metcalf external cavities 
were used for the two-step excitation scheme. The first (X,i) laser step (6 2 S 1/2 -> 6 2 ?° ia ) 
was provided by a diode laser (Model 2010A, Newport Corp., USA) with an output 
power of 16 mW at 852.12 run and a linewidth of approximately 5 MHz. Figure 4-5 
shows the experimental setup where the \ l laser passed through an amorphic prism pair 



59 

to shape and expand the beam to approximately 4 mm x 10 mm. The expanded beam then 
passed through a transmission target (USAF 1951 3-Bar Resolving Power Test Target). A 
cesium/neon hollow cathode lamp (Model 14-386-100J, Fisher Scientific, USA) and a 
silver/argon hollow cathode lamp (Model 45-483, Jarrell Ash, USA) were also used as X] 
photon sources. In either case, the cathode of the lamp was used as the image source. The 
resulting X] image was then passed through a dielectric mirror positioned at 45° relative 
to the input window. The dielectric mirror transmitted near-infrared and reflected visible. 
The image then passed into a cylindrical pyrex RFIM cell (Opthos Instruments, 
Rockville, MD, USA), 75 mm in length and 25 mm in diameter, containing metallic 
cesium under a vacuum. Cesium atoms absorbing ^i photons are further excited within 
the cell by the output of a second diode laser (Model TEC-500, Sacher Lasertechnik, 
Marburg, Germany). This laser (k 2 ) was used as the pump laser at 917.23 nm (6 2 P 3 / 2 -> 
6 D 5/2 ) and had an output power of 30 mW and a linewidth of 5 MHz. 

In the orthogonal beam geometry described previously [12], an imaging plane, 
defined by the shape and placement of the X 2 laser, defined the cesium/photon interaction 
region. The imaging plane was a sheet of light 12 mm long and 0.5 mm wide created 
using a pair of cylindrical lenses to transform the output of the X 2 laser. This sheet of light 
was directed near the input window orthogonal to the length of the cell. In the co- 
propagating beam geometry, the X 2 laser output was expanded with a beam expander or 
lens and directed into the cell through the input window. This was accomplished by 
directing the X 2 laser towards a beam splitter in the X\ path. 



60 



CL^ 



Cs Cell 




CCD 

Camera 



Figure 4-5. Experimental setup for the RFIM. M = Mirror, DM = Dielectric Mirror, APP 
= Amorphic Prism Pair, NDF = Neutral Density Filter, SPF = Short Pass Filter, CL = 
Cylindrical Lens, L = Lens, I = Image Intensifier, BS = Beam Splitter, BE = Beam 
Expander. 



61 

Atoms excited through absorption of both X\ and X 2 photons can radiatively decay 
to the 7 2 P°3/ 2 state through a radiative transition (14.35 urn, 6 2 D 5/2 ■> 7 2 P 3 / 2 ) and then 
fluoresce to the ground state (7 2 P 3 / 2 -» 6 2 Si /2 ) emitting photons at 455.53 nm. This blue 
fluorescence was reflected at 90° by the dielectric mirror, passed through a 650 nm 
shortpass filter and was imaged onto the photocathode surface of a single-stage, 
proximity focused image intensifier (Model V807OU-64-N132, Hamamatsu Corporation, 
Japan) with a gain of 10 4 . Photons emitted from the phosphor-coated back surface were 
imaged onto a cooled, digital CCD camera (Model Penguin 150CL, Pixera Corporation, 
Los Gatos, CA) that was computer-controlled. The CCD camera software allowed for 
image integration times from 0.1 ms to 60 s and also provided image averaging and 
summing capabilities. For a given Xj power, the signal-to-noise increased with 
integration time; as a result, multiple images are typically acquired to determine the 
maximum signal-to-noise (largest integration time) while remaining beneath the 
saturation threshold for the CCD pixels. Images were further processed using the 
SigmaScan Pro software package (Version 5.0, SPSS, Chicago, IL, USA). Such software 
processing included intensity enhancements, such as gamma correction, to visualize 
especially faint images. The system (cesium cell, filters, and image intensifier) was 
enclosed in a black plexiglass box to minimize stray light incident on the image 
intensifier photocathode and the CCD camera. Both the signal (fcj) photons and the pump 
(A, 2 ) laser entered the enclosure through windows fitted with 750 nm longpass filters. 
Results 

The spectral response of the RFIM was studied using two hollow cathode lamps. 
Several images were acquired using a cesium hollow cathode lamp as the A,, photon 



62 

source, while the \q pump laser was oriented in the orthogonal or co-propagating 
geometry. The lamp was positioned such that a 1 : 1 image of the cathode was incident 
upon the cesium vacuum cell. The lamp had a recommended maximum operating current 
of 25 mA. Faint images of the cathode could be seen at 5 mA, while distinct images 
could be seen above 10 mA using an integration time of 30 s. An image of the D 2 
emission (852.12 nm) from the cesium lamp (25 mA) using orthogonal beam geometry is 
shown in Figure 4-6. Images were also acquired using a silver/argon hollow cathode 
lamp (25 mA) in order to demonstrate the spectral background rejection of the RFIM. 
The argon emission at 852.144 nm (4S*i/2 -> 4P 3/2 ) was eleven times more intense than 
the cesium emission. The argon line is located 12 GHz (29 pm) away from the center of 
the cesium absorption line. As expected, no image was detected using the silver/argon 
hollow cathode lamp since the argon emission line was outside of the narrow RFIM 
spectral response. Orthogonal and co-propagating geometries performed identically, 
which is expected since the 12 GHz shift is outside the absorption band for either 
orientation of detection. 

To evaluate the spectral resolution of the RFIM, images were acquired using the 
transmission bar target and the X\ laser tuned to excite atoms from either the F = 3 or F = 

4 hyperfine level of the ground state. The X 2 laser was tuned to excite atoms from the F = 

5 hyperfine level of the 6 2 P 3 / 2 state. Only the atoms originating from the F = 4 level of 
the ground state will be excited by the X 2 laser because the F = 3 (6 2 S ]/2 ) to F = 5 (6 2 ?° m ) 
transition is forbidden. Figure 4-7 shows a bright image when X\ is tuned to the F = 4 
level, while no image is seen when tuned to the F = 3 level. Only one of the cesium 



63 



Figure 4-6. Image of the D 2 emission from a cesium hollow cathode lamp (HCL). The 
image shown is a 1:1 image of the 4mm diameter circular cathode. No image was seen 
from the argon emission line (852.14 nm, 4S°i /2 -> 4P 3/2 ) of a silver/argon HCL. 



64 

RFIM hyperfine "channels" produced a signal, and the overall spectral resolution of the 
Rf IM is limited by the F= 4 -> F = 5 hyperfine Doppler linewidth (400 MHz) when 
using orthogonal beam geometry. It should be noted that only one hyperfine channel is 
detected in the orthogonal geometry as well. 

In order to directly compare the spectral resolution in the two beam geometries, a 
series of images of the X\ laser through the transmission mask was taken in both 
orthogonal and co-propagating beam geometries at various frequency shifts from the 
center of the cesium D 2 absorption line. Figure 4-8a is a representative image of the Xi 
laser when detuned by a 230 MHz shift (achieved by deflecting a piezoelectric element 
placed behind the external cavity tuning mirror) when in orthogonal beam geometry, 
while only a faint image (Figure 4-8b) is seen in co-propagating geometry at a 220 MHz 
shift. The images acquired in orthogonal geometry are consistent with the results 
obtained with the RFM: the spectral response of the detector was limited by the Doppler 
broadened profile at the cell temperature (400 MHz at room temperature); however, when 
the beams entered the RFIM in co-propagating geometry, the spectral response became 
narrower than the Doppler broadened profile. The two beams in this arrangement excited 
only the atoms within a certain velocity subset, which allowed sub-Doppler resolution to 
be achieved. Based on pixel intensity as a function of Vj frequency, the FWHM of the 
detector response is 270 MHz when the beams are in the co-propagating geometry. 



65 




200 jum 



Figure 4-7. Image acquired of the transmission bar target using the 3l, laser tuned to 
excite atoms from either the F = 4 hyperfine level of the ground state. No image was 
seen when tuned to excite the F = 3 hyperfine level. 



66 



a) 



b) 




Figure 4-8. Images of the X\ (852.12 nm) laser through the transmission mask A) while 
h is tuned 230 MHz away from the center of the absorption profile. An image is 
obtained because this shift is within the spectral response of the orthogonal RFM. B) 
Image acquired at a 220 MHz shift from the center of the absorption band in co- 
propagating geometry. The image is faint and is only discernable using software 
enhancements. In co-propagating geometry, a 220 MHz corresponds to the wings of the 
absorption profile. 



67 

Conclusions 

The observation of sub-Doppler spectral resolution in non-imaging and imaging 
resonance fluorescence detectors has been demonstrated. This spectral resolution is 
possible due to selective excitation of specific atomic populations. Excitation profiles 
obtained using the RFM indicate a spectral resolution on the order of 200 MHz. This 
instrument response is tunable under the Doppler broadened profile by manually 
changing the X 2 (917 nm) laser wavelength. 

We report a similar spectral resolution (270 MHz) in the RFIM. The spectral 
response of the RFIM is somewhat broader than in the RFM. This discrepancy could be 
explained by varying degrees of overlaps of the expanded beams entering the RFIM. In 
both the RFM and RFIM, care was taken to ensure that the beams were overlapping; 
however, it was more difficult to overlap the two expanded beams using the RFIM. In 
addition, any inconsistencies with the beam expansion could create portions of the 
expanded beam that are not collinear. 

Images were obtained of a cesium hollow cathode lamp and compared to those 
obtained of a silver/argon hollow cathode lamp. Images were observed only using the 
cesium lamp. The images obtained serve as an independent demonstration of the spectral 
resolving capabilities of the RFIM. Likewise, images were obtained where \, (852 nm) 
was tuned to excite atoms originated from either the F = 4 or F = 3 ground state. When 
the X 2 tuning position was chosen appropriately, only excitations originating from the F - 
4 state were observed. Further evaluations of spectral resolution of the RFM and RFIM 
are made in the next chapter, where moving objects are detected by the Doppler-shift that 



68 

occurs when the moving object was illuminated with the A-i laser source. Investigations 
are made using a using a rotating disc and cesium, sub-Doppler RFM and RFIM. 



CHAPTER 5 
MOVING OBJECT DETECTION AND IMAGING 

Introduction 

Atomic vapor filters have been applied towards a variety of laser Doppler 
velocimetry (LDV) techniques in recent years. Such filters provide high spectral 
resolution resulting from the selective absorption of resonance photons in an atomic 
vapor. This spectral resolution is necessary to spectrally distinguish the Doppler-shifted 
photons (resulting from the illumination of a moving object with laser radiation) from the 
photons incident upon the moving object. The maximum Doppler frequency shift of 

scattered laser radiation from a moving object with velocity, V, can be described by 

2V 
Av = — ! where A is the wavelength of incident laser radiation. This maximum 

X 

shift represents the case where the object's velocity vector is directed toward the 
illumination source and also toward the detector of the scatter. In other cases, the factor 
of 2 shown in Equation 1 will be replaced by a value between and 2, dependent upon 
the angle between illumination and detection [41]. 

Several LDVs have been reported in the literature based on one-step absorption 
and fluorescence in an atomic vapor [42-45]; such schemes are limited by Doppler- 
broadening in the atomic vapor. Bloom et al, reported on an LDV based on cesium 
absorption [42,43]. The Doppler-shifted photons were detected by utilizing sharp changes 
at the edge of the absorption profile. Lempert et al, reported on an LDV technique 
combining frequency-modulated absorption spectroscopy and absorption of resonance 



69 



70 

photons by a potassium-vapor cell [44,45]. We report on an LDV based on a two-step 
excitation scheme with sub-Doppler spectral resolution. The cesium resonance 
fluorescence monochromator (RFM) reported here is based on selective absorption of 
photons (X.1 = 852.12 nm) corresponding to a ground-state transition (6 2 Si/ 2 -> 6 2 P°3/2) 
followed by laser pumped excitation to the 6 2 D 5/2 level at 917.23 nm (k 2 ). The initial 
absorption process can be detected by monitoring the wavelength-shifted fluorescence at 
455.53 nm (7 2 P°3/2 "> 6 2 Si/ 2 ) along the decay pathway from the 6 2 D 5/2 state (6 2 D 5/2 
-*7 2 P°3 /2 -> 6 Si /2 ). We have recently reported sub-Doppler spectral resolution (200 
MHz FWHM) through excitation of a monokinetic population of atoms [31]. We now 
report on the application of this RFM to moving object detection. In addition, we also 
demonstrate the imaging of Doppler-shifted photons using the cesium resonance 
fluorescence imaging monochromator (RFIM). 

Moving Object Detection Using a Resonance Fluorescence Detector 
Experimental Setup 

A rotating disc was used to generate the desired Doppler shift in X| as this laser 

was scanned in wavelength. The experimental setup is shown in Figure 5-1. The incident 

light, Xi + AX, was provided by a tunable, external cavity diode laser (Model 2010A, 

Newport, Irvine, California). This wavelength was scanned by applying a waveform (3 

Hz sawtooth) from a function generator (Model F3GB, Wavetek, San Diego, California) 

to a piezoelectric element behind the Littman-Metcalf cavity. The lasing output was 

directed toward an aluminum, circular disc (5 cm diameter, 0.7 cm thick) that was rotated 

(between 135 Hz and 417 Hz) by attaching it to a modified motor from a commercially 

available router power tool (Model DW610, DeWalt, Baltimore, Maryland). Two notches 



71 



852 nm 
Laser 



Function 
Generator 



Removable 
Mirror 



HeNe 
Laser 



Collection 

J. 



917 nm 
Laser 



:ns 



PMT^* 



& 



* i Lens 
Microscope Slide 

Interference 
/ 4r /Tilter 



V/A 



Oscilloscope 



I 



Boxcar 
Integrator 




Rotating 
Disc 



Photodiode 



o 



V/A 



Figure 5-1. Experimental setup showing detection of Doppler-shifts using the cesium 
RFM. The 852 nm (k\) laser is frequency scanned and illuminates the rotating disc. The 
scatter is directed into the cesium cell where absorption of Xi and 917 nm (k 2 ) photons 
results in wavelength-shifted fluorescence at 455 nm. Excitation profiles are generated by 
the boxcar integrator, which is triggered by the waveform that scans the Xi laser. In 
addition, a slight divergence between the two laser beams prevents unwanted reflections 
from entering either laser cavity. 



72 

were cut into the disc on opposing sides to provide a surface for illumination. A thin 
mirror was attached to each notch to increase the amount of light reflected from the 
moving disc. The area of each notch was 1 cm by 0.7 cm. The disc was rotated such that 
the notches were moving toward the incident light thus maximizing the Doppler shift; for 
the same reason, the angle between the incident light and the collection optics was 
minimized. The scattered light was collected and focused into the RFM by a 16 cm focal 
length lens. The cylindrical cesium cell (Opthos Instruments, Rockville, Maryland) was 
made of pyrex and had dimensions 75 mm by 25 mm; this cell contained a fill of solid 
cesium metal under a vacuum. The scattered radiation was directed into the front window 
of the cell. A second external cavity diode laser (Model TEC500, Sacher Lasertechnik, 
Marburg, Germany) was tuned to the 6 2 P 3 / 2 -> 6 2 D 5/2 transition (X 2 = 917.23 nra) of 
cesium and directed through the back of the cell; thus the beams were in counter- 
propagating geometry. When a X x photon was absorbed and subsequently excited by X 2 , 
the resulting fluorescence from the 7 2 P° 3 /2 -> 6 2 Si/2 transition (455.53 nm) was collected 
through the side of the cell by a lens and focused through a 455 nm interference filter (10 
nm FWHM) onto the surface of a photomultiplier tube (Model 928, Hamamatsu, Japan). 
The output from the photomultiplier tube was passed through a current amplifier (Model 
SR570, Stanford Research Systems, Sunnyvale, California) and sampled by boxcar 
integrator (Model SR250, Stanford Research Systems, Sunnyvale, California). The 
boxcar was triggered by a photodiode placed near the disc, using a HeNe laser placed on 
the opposing side of the disc. The output from the boxcar was connected to an 
oscilloscope that was triggered by the waveform provided by the function generator. This 
method allowed for intensity measurements to be made as a function of X] tuning 



73 

position. A removable mirror and a microscope slide were placed in the X\ beam path and 

provided a reference (at rest) measurement of the excitation profile. 

Results 

A representative velocity measurement is shown in Figure 5-2. The smaller peak 
shown is of the reference beam; the relatively small size of this peak is due to the 
microscope slide directing only a fraction of the beam into the cesium cell. The larger 
peak represents scattered light from the disc when moving at 66 m/s (0.05 cm radius, 210 
Hz). These peaks were made up of small step-functions, each step representing one of the 
disc notches aligned for proper interaction with the X,i laser. Averaging 16 acquisitions on 
the oscilloscope smoothed the peaks in order to determine the frequency location of the 
maximum peak fluorescence. The frequency difference between reference and Doppler- 
shifted peaks shown in Figure 5-2 was 142 MHz, which when applied to Equation 1, 
corresponds to a velocity of 60 m/s. This value is in good agreement with the velocity 
measurement (66 +/- 7 m/s) calculated using the frequency of the trigger signal and the 
radius of interaction. 

A series of traces was obtained at fixed velocities to determine the fluctuation in 
the acquired X\ traces. Twenty acquisitions were made at 66 m/s and the uncertainty in 
this measurement process was determined to be on the order of +/- 9 m/s. Given this 
uncertainty, we should be able to distinguish velocities on the order of 27 m/s. Two 
different velocity measurements of this magnitude, along with the reference, are shown in 
Figure 5-3. Short-term fluctuations during the acquisition and averaging process are 
expected to contribute to the uncertainty, while long-term fluctuations do not contribute 
to the uncertainty, since this drift would equally affect both the reference and shifted 



74 





l^t - 






12 




^_^ 






d 


10 




ro 






■"*— * 


- 




>? 






CO 


8 




c 






cd 






-t-> 






_c 








6- 




CD 






O 






c 






CD 






O 


4- 




V) 






CD 






i_ 


- 




O 






^ 


2 




Li- 


0^ 









60m/s 
Reference 




-2000 -1500 -1000 -500 

Frequency (MHz) 



T 




500 



Figure 5-2. Excitation profiles of reference beam and scatter from rotating disc. The 
frequency difference (142 MHz) between the two peaks represents a velocity of 60 m/s. 



75 



9-, 



7- 



3 6 
t 5 

Q 

<D 
-*— < 

a 4-1 
C -i 

BO 

o 



96m/s 
121 m/s 
Reference 




i ■ 1 ' 1 " 1 

-2000 -1500 -1000 -500 

Frequency (MHz) 







500 



Figure 5-3. Excitation profiles of reference beam and scatter representing two different 
disc velocities. 






76 

peaks. When multiple acquisitions were obtained of the reference, the corresponding 

uncertainty was +/- 5 m/s, which is less than the value obtained using the Doppler-shifted 

peak. To account for this discrepancy, variations in the rotation of the disc were 

monitored as a function of time using the HeNe/photodiode trigger. At a fixed velocity 

setting, the time between trigger events varied by as much as +/- 10 % of the average 

value. The variations in the disc rotation can contribute to the larger uncertainty in 

measuring the position of the Doppler-shifted peak. 

Ultimately, the limiting sources of uncertainty in such measurements are 

fluctuations in laser frequency and the linewidths of the lasers themselves (5 MHz). 

Based upon short-term fluctuations in the reference peak, distinguishing minimum 

frequency shifts on the order of 36 MHz (15 m/s) can be achieved, which is a factor of 

three larger than the uncertainty in the frequency obtained by a typical measurement. Just 

as the minimum frequency shift dictates the minimum detectable change in velocity, the 

maximum frequency shift defines the upper limit of velocity detection. This figure of 

merit results from the tuning capabilities of the laser that interacts with the moving 

object. In these studies, the Xi diode laser was scanned over a 3.5 GHz region. If the scan 

is chosen such that the reference peak resides near the edge of this scan, a working 

detection range from m/s to 1500 m/s can be achieved. 

Imaging Doppler-shifted photons 
Experimental Setup 

Similar to the studies performed using the single-point RFM, a rotating disc was 

used in these studies to generate the Doppler-shift in X\ (see Figure 5-4). In this case, 

however, the A,] laser was not scanned in wavelength using a function generator. Instead, 



X 



77 



I 



BE 




RFIM 



X 



Figure 5-4. Experimental setup showing imaging of Doppler-shifted photons using the 
cesium RFIM in co-propagating geometry. 



78 

the X] laser was manually tuned to fixed positions. The light incident on the disc, X { + &k, 
was provided by an external cavity diode laser (Model 2010A, Newport, Irvine, 
California). The disc was aluminum and circular (5 cm diameter, 1 .0 cm thick) and was 
rotated (between 135 Hz and 417 Hz) by attaching it to a modified motor from a 
commercially available router power tool (Model DW610, DeWalt, Baltimore, 
Maryland). The disc used had four notches, and the X] scatter was directed into the RFIM 
when a notch was in the correct orientation. The scatter was collected by a lens and 
combined with the X 2 beam, which was expanded using a beam expander. Both beams 
entered the RFIM in co-propagating geometry. The X 2 beam was provided by a second 
external cavity diode laser (Model TEC500, Sacher Lasertechnik, Marburg, Germany) 
which was tuned to the 6 2 P°3 /2 -> 6 2 D 5/2 transition (X 2 - 917.23 run) of cesium. The two 
beams entering the RFIM passed through a dielectric mirror (which transmits near- 
infrared and reflects visible) and into a cylindrical pyrex RFIM cell (Opthos Instruments, 
Rockville, Maryland), 75 mm in length and 25 mm in diameter, containing metallic 
cesium under a vacuum. Cesium atoms absorbing both Xi and X 2 photons can fluoresce to 
the ground state (7 2 P 3 / 2 -> 6 2 S 1/2 ) emitting photons at 455.53 nm. This fluorescence was 
reflected by the dielectric mirror, passed through a 650 nm shortpass filter and imaged 
onto the photocathode surface of a single-stage, proximity focused image intensifier 
(Model V807OU-64-N132, Hamamatsu Corporation, Japan) with a gain of 10 4 . Photons 
emitted from the phosphor-coated back surface were imaged onto a cooled, digital CCD 
camera (Model Penguin 150CL, Pixera Corporation, Los Gatos, CA) that was computer- 
controlled. Typical integration times were between 0.5 and 10 seconds. Images were 
further processed using the SigmaScan Pro software package (Version 5.0, SPSS, 



79 

Chicago, IL). The system (cesium cell, filters, and image intensifier) was enclosed in a 
black plexiglass box to minimize stray light incident on the image intensifier 
photocathode and the CCD camera. The X { and X 2 photons entered the enclosure through 
a window fitted with a 750 nm longpass filter. 
Results 

An image of the scatter from the disc surface when the disc was at rest is shown 
in Figure 5-5. This image was taken with a 0.5 second integration time and represents the 
image obtained when both X| and X 2 lasers are tuned to their respective absorption 
maximum of the cesium transitions. When the disc was spun at 383 Hz, no image was 
observed in this tuning position regardless of the integration time (10 s, 20 s, and 60 s 
were attempted). At a disc radius of 5 cm, the velocity at this rotational frequency is 120 
m/s. This velocity corresponds to a Doppler-shift of 282 MHz. Since the half-width at 
half-maximum (HWHM) is on the order of 135 MHz, photons shifted 282 MHz away 
from resonance would not be expected to generate an image. To create images of the 
Doppler-shifted photons, images were acquired while the \\ laser was incrementally 
tuned away from resonance. When the k\ laser was tuned away from resonance by 150 
MHz, an image was observed (10 s integration time) and is shown in Figure 5-6. 
Although the expected Doppler-shift is 282 MHz away from the absorption maximum, an 
image can be observed when tuned by a lesser value since the absorption profile has a 
given width. To determine the velocity of the disc using the RFIM, the X| laser was tuned 
until the most intense image was observed. The image shown in Figure 5-7 is the most 
intense image (10 s integration time) obtained and occurred at a shift on the order of 300 
MHz. 



80 



Figure 5-5. An image of scatter from the disc surface when the disc was at rest. Both 
lasers were tuned to the absorption maximum, and the integration time was 0.5 seconds. 



81 




Figure 5-6. Image obtained when the X t laser was detuned by 150 MHz and disc velocity 
at 5 cm was 120 m/s. A 10 second integration time was used. The image shown has been 
enhanced for reproduction. 



82 




Figure 5-7. Image obtained when the X { laser was detuned by 300 MHz and disc velocity 
at 5 cm was 120 m/s. This tuning position resulted in images with the highest intensity. 
The integration time was 10 seconds. This image has been enhanced by the same degree 
as the image shown in Figure 5-6. 



83 

Point-spread functions taken of the three images are shown in Figure 5-8. The image 
obtained using a 300 MHz shift had a larger intensity than that obtained at a 150 MHz 
shift. This figure also shows how no image is obtained when the X\ laser is tuned to 
resonance (0 MHz shift). 

In order to determine the smallest frequency shift that can be detected with the 
PvFIM, a series of images were obtained while the disc was at rest. The rationale was to 
quantify differences between subsequent images taken under the same conditions. To this 
end, seven images were taken at an integration time of 0.5 seconds. Point-spread 
functions for these images are shown in Figure 5-9. Specific pixels were chosen and the 
intensity was compared to the corresponding pixels in the other images obtained. The 
variation in the pixels intensity was consistently between +/- 4 and +/- 7, which suggests 
that images can be discerned when differences of 20 (pixel intensity units) are observed. 
The variation in pixel intensities can lead to uncertainties in obtaining accurate velocity 
measurements. The relationship between changes in pixel intensity and changes in 
frequency (MHz) depends on the particular location along the tuning profile, but can be 
on the order of 20 MHz to 30 MHz. The uncertainty can be decreased, however, when a 
clusters of pixels are considered. When pixels in the same vicinity were averaged, the 
variation in intensity between subsequent images was as low as +/- 1 . 



84 



100 
90 
80 



300 MHz shift 




150 MHz shift 
MHz shift 



150 200 

Pixel number 



250 



300 



350 



Figure 5-8. Point-spread functions of the images acquired at frequency shifts of MHz 
150 MHz, and 300 MHz. 



85 



1/3 



120 -| 

100- 

80- 

60- 



X 



20 



0- 



-50 








— r~ 
50 



100 150 200 250 
Pixel number 



300 350 400 



Figure 5-9. Point-spread functions of seven subsequent images acquired when the disc 
was at rest at 0.5 second integration time. 



86 

Conclusions 

In conclusion, a two-step, cesium resonance fluorescence monochromator with 
sub-Doppler spectral resolution has been evaluated for use as a laser Doppler 
velocimeter. A velocity range of m/s to 1500 m/s and a limiting velocity change of 25 
m/s have been realized using the current experimental setup. These values can be 
improved through further improvements in spectral resolution. In the current study, each 
data point on the laser frequency scan represented an individual detection event as the 
disc was rotated. In other words, each rotation of the two-notch disc generated two data 
points on a given laser frequency scan. One possible improvement is the use of a single, 
rapid laser scan to detect Doppler shifts in a single detection event. Such improvements 
would decrease the complexity of the detection while increasing the ability to distinguish 
small velocity changes; however, it is important to note that in this experimental design, 
the interaction time between the laser and the moving object must be long enough to 
achieve a single frequency scan. 

The detection of moving objects has also been demonstrated by images acquired 
by the cesium RFIM. Images of Doppler-shifted (282 MHz) photons were acquired; 
based on image intensities, the magnitude of the Doppler-shift was approximately 300 
MHz. This value is reasonable when considering the possible drift in either diode laser 
over the relatively large integration times (10 seconds). The time delay between 
subsequent acquisitions also contributed to this effect. This delay resulted from a 
combination of two factors. The first is the short time required to manually tune the X,, 
laser to a new position. A much larger delay is due to the image acquisition software; the 



87 

software must process and save a given acquisition before subsequent acquisitions can be 
made. 



CHAPTER 6 
CONCLUSION 

The scope of this work can be summarized by four main points. First, two 
ionization schemes and one fluorescence scheme were evaluated experimentally. The 
highest quantum efficiency (1%) was found for the three-step ionization scheme, which is 
due to the high irradiance provided by the Nd:YAG laser used to accomplish ionization. 
Although fluorescence detection was less efficient, it was used in subsequent studies due 
to its relative ease of implementation and also due to difficulties in constructing a cesium 
ionization imaging detector. 

Second, resonance fluorescence detection was evaluated theoretically in terms of 
quantum efficiency and in the identification of competing pathways. A rate-equation 
model was developed using MathCad (MathSoft, Inc.U.S.A.); additionally, a model 
consisting of fewer atomic levels was investigated using the density matrix formalism. 
The 7 P°3/ 2 •> 6 S1/2 (455.53 nm) was identified as the pathway with the largest 
fluorescence intensity with smaller contributions from other radiative transitions. 

Third, a method was devised to improve the spectral resolution of non-imaging 
and imaging photon detection. The spectral resolution has previously been limited by the 
Doppler-broadened absorption linewidth, but in these studies specific populations of 
atoms were excited through alignment of laser beam geometries. The spectral resolution 
achieved using a cesium RFM (resonance fluorescence monochromator) and RFIM 
(resonance fluorescence imaging monochromator) was 200 MHz and 270 MHz, 
respectively. 



88 



89 

Fourth, additional efforts demonstrated the use of the cesium RFM and RFIM 
toward moving object detection. In this last study, a rotating disc was used to generate 
Doppler-shifted photons, which were detected using the RFM and also imaged using the 
RFIM. A key difference between the two experimental designs is that the RFM is a time- 
gated detector that detected photons when the disc was in the correct orientation. In 
addition, the RFM was frequency scanned using a function generator; this allowed for 
entire spectral profiles to be created and compared to a reference. The RFIM acquires 
images based on a variable integration time, so the RFIM integrated images even when 
the notches were not in the correct orientation for interaction with X|. Even though the 
use of the rotating disc resulting in a small duty cycle, images of Doppler-shifted photons 
were successfully obtained. The RFIM, as used in this work, is best suited for the 
detection of moving objects that do not change temporally or spatially over the required 
integration time. One example of such application is the imaging of air or particulate 
flows under conditions that remain static throughout the duration of an acquisition. 

The significance of this work is fourfold. The evaluation of the two ionization 
schemes in this work serve as an incremental step in the eventual development of a 
cesium resonance ionization imaging detector. A cesium RIID will likely involve pulsed 
excitation and ionization in order to achieve a high efficiency. 

The evaluation of the fluorescence detection scheme adds to the understanding of 
the cesium RFIM developed in our laboratory. The theoretical studies of the fluorescence 
scheme demonstrated that a model can be useful in choosing a scheme, since identifying 
competing pathways is possible. Regardless of the complexity of a proposed model, such 



90 

theoretical studies can be useful in evaluating fluorescence detection schemes and also 
ionization schemes. 

The selective excitation of cesium atoms has been shown to narrow the spectral 
response of photon detection and imaging. The spectral resolution is a key figure of merit 
for resonance fluorescence and ionization detectors and imaging systems. The 
improvements realized in this work are incremental when considering that improvements 
on the order of 50% were realized. More significant than the actual values obtained is that 
several other techniques (to selectively excite atoms) described in the literature can be 
applied to photon detection and imaging. Examples of such techniques are described in 
the following chapter. 

The successful demonstration of these photon detectors and imaging systems 
towards moving object detection is a first step in their development towards true 
application. Although further improvements are needed before application, the detection 
and imaging of Doppler-shifted photons described in this work can provide a starting 
point for future work. 






CHAPTER 7 
FUTURE WORK 

Photon detection and imaging by atomic vapor based photon detectors remain in 
the relative early stages of development. Future studies in this field will be discussed in 
the context of cesium chemistry and spectroscopy, although these techniques may apply 
to other suitable elements. The improvements for this technology can be divided into two 
categories: sensitivity and spectral selectivity. 

The task of improving the sensitivity of cesium based photon detectors can be 
approached from multiple directions. In the context of resonance fluorescence detection 
and imaging, the sensitivity is limited by the chosen excitation scheme and the fraction of 
atoms that fluoresce down a particular pathway. Improvements in the collection 
efficiency of the fluorescence signal would improve the capabilities of these detectors. 
Further developments in cesium ionization could result in improved sensitivity. In this 
work, direct photoionization was evaluated; however, other forms of ionization can be 
investigated in future studies, namely collisional ionization [26,46,47] and field 
ionization [26,48]. In addition to optimizing the spectroscopy involved in developing 
ionization detectors, an additional task is the successful development of cesium ionization 
imaging cell. 

A variety of methods could be employed to improve the spectral resolution of 
photon detection. Once developed, these methods could be applied to either fluorescence 
or ionization detection. Saturation spectroscopy was investigated in this work [33-40], 
which was the first demonstration of sub-Doppler spectral resolution in atomic vapor 

91 



92 

based detectors. In all previous studies, the spectral resolution was limited by Doppler 
broadening in the atomic vapor. In addition to saturation spectroscopy, a variety of other 
techniques in high-resolution spectroscopy could be applied to the field of high- 
resolution photon detection. One such example is atomic beam spectroscopy [33,49,50]. 
Since the atoms share a common velocity in an atomic beam, the process of absorption 
will become selective since the velocity distribution will be much narrower than the 
Doppler-broadened profile. Essentially, the Gaussian distribution of atomic velocities is 
replaced by a population of atoms that will absorb photons in a narrow spectral range. 

A second method to improve the spectral resolution also limits the velocity 
distribution of atomic velocities. The atomic velocities within a cesium cell could be 
limited if a thin cell were constructed. Such cells have been used previously in the field 
of cesium spectroscopy [51-54]. Such cells are typically 1 - 10 mm thick; however, cells 
as thin as 150 nm have been reported [51]. 

The spectral resolution of photon detection could also be improved by 
introduction of an additional laser-pumped transition that is coupled to ground state. In 
this arrangement, the ground state atoms would either absorb photons via the detection 
pathway, or these atoms would be excited via the competing transition. The competing 
transition could excite selected populations of atoms, which would result in sub-Doppler 
spectral resolution in the detection pathway. Techniques that operate in this manner 
include population depletion and Rabi splitting [55]. 

Electromagnetically induced transparency (EIT) also occurs via the coupling of 
common atomic states [55-61]. In EIT, a strong laser beam, called the pump beam, is 
used to excite a given transition, while a weaker, probe beam is oftentimes wavelength 



93 

tuned along a second transition. Although several variations exist, in all cases the two 
transitions share a common level. In the context of photon detection, the coupled level 
will likely be the ground state. Figure 7-1 shows an energy level diagram of two-step 
excitation for resonance fluorescence detection with a pump laser coupled to the ground 
state. When the pump laser is turned off, the output of the probe will be entirely absorbed 
by the atoms if the probe is weak. When the pump laser is on, a spectrally narrow portion 
of light is transmitted through this cell. Under specific cicrumstances, a linewidth of 42 
Hz has been reported for cesium in the literature [62]; it is worth noting that this value is 
less than the natural linewidth. This phenomenon is explained as a quantum interference 
between the coupled transitions. A cesium cell under EIT conditions would transmit a 
spectrally narrow window of light; this cell could be placed in front of a cesium 
resonance fluorescence or ionization detector and serve as a prefilter of the incoming 
photons. Essentially, the first cell will absorb all photons within the Doppler-broadened 
profile except for those falling within the narrow, transmitted region. The effective 
spectral resolution would then be defined by the spectral profile of the transmission 
through this first cell. The spectral resolution provided by EIT would likely be the 
ultimate in spectral resolution. As the EIT effect is increased (narrower FWHM), the 
transmission through the EIT cell decreases. The ideal EIT conditions would likely be a 
compromise between spectral resolution and throughput. 

In the fifth chapter of this work, demonstration of moving object detection was 
described. Through a combination of improved sensitivity and spectral resolution, this 
technology could be applied to a variety of applications. One such application is the 
measurements of air or particulate flows in field of study such as aerodynamics and the 



94 

analysis of combustion processes. Depending on the intensity and coherence of the signal 
laser source, remote-sensing applications are also possibilities. With significant 
improvements in spectral resolution and sensitivity, biological motion detection, such as 
blood flow analysis, may also be possible. Once further developments and improvements 
are in place, implementing these cesium-based photon detectors for true applications will 
likely be the goal of future projects. 



7 2 P°" 

1 r 3/2 



455.53 nm 



V* 



35^ 



6 2 P° 

° r l/2 



^ P ump = 894 - 3 5 n n 



V 



7K — 6 ° 5/2 



917.23 nm 




Figure 7-1. Energy level diagram showing two-step excitation with a pump laser coupled 

tO the SrOUnd-State. In this fai!S tVlP nnmn lacm- nnprat^ ot CO/I ?< „~, ♦« <,„„.*„ *U_ Ac 



to the ground-state. In this case, the pump laser operates at 894.35 nm to excite the 6^S 



-> 6~P°i/2 transition 



1/2 



APPENDIX 
RATE EQUATION MODEL OF EIGHT CESIUM LEVELS 

The rate equation model shown below was created using MathCad (MathSoft, 

Inc., U.S.A.). Eight rate equations were produced based on pumping rates and 

spontaneous emission processes. Four of the levels are directly involved in the cesium 

resonance fluorescence detection scheme, while the other four levels can be populated 

through competing pathways of radiative decay. 

Comments throughout the MathCad worksheet are shown in italic font. 



Atomic 
Level 


Population 
Fraction 


6 S1/2 


Ql 


6 2 P°3/2 


n 2 


6'D 5 /2 


n3 


V 2 P°3/2 


ru 


7 2 S 1/2 


n 5 


5 2 D 5/2 


n 6 


5*Dm 


n? 


6'P°,/2 


n 8 



Each variable, n x , represents the population fraction 
of an atomic level, x. 

Variables are defined that represent the rates populating 
and depleting the various levels. The labeling scheme is 
consistent with that shown above; the variable A xy represents 
the rate of spontaneous emission from level x to level y. 

Each variable B xy represents the product of the absorption cross section, a (cm 2 ) and 
the irradiance (photons W cm' 2 ). 



95 



96 



A 21 


:=3.70-10' 


B 21 


= 1.145610 5 


B 12 


= 5.728 MO 4 


A 32 


:=1.74-10 7 


B 23 


= 2.47-10 6 


B 32 


= 1.24-10 6 


A 34 


= 6.6710 4 


A 31 


= 2.0210' 


A 41 


= 4.18610 6 


A 45 


= 3.9810 6 


A 46 


= 8.35-10 5 


A 47 


= 7.52-10 4 


A 52 


= 1.22-10 7 


A 58 


= 5.5310 6 


A 61 


= 2.07-10 6 


A 62 


= 8.4-10 5 


A 72 


= 1.22-10 5 


A 78 


= 1.0-10 6 


A 71 : 


= 1.96-10" 1 


A 81 : 


= 2.89-10 7 



Initial values assigned to initialize the iterative process. 

n j :=0.20 
n 2 :=0.20 
n 3 :=0.20 
n 4 :=0.12 
n 5 :=0.12 
n 6 :=0.12 
n ? :=0.12 
n fi :=0.12 



97 



Initial constraints on the solution set. Further restrictions can be used if MinErr provides 
erroneous solutions. 

Given 



n j<1 


n 2 < ' 


n 3 <l 


n 4 <l 


n 5 <1 


n 6 <l 


n 7 <l 


n g <l 


n j>0 


n 2 >0 


n 3 >0 


n 4 >0 


n 5 >0 


n 6 >0 


n y>0 


n 8 >0 



98 

Nine equations: 

B 12 )-n 1+ (B 21 + A 21 ).n 2+ (A3 1 ).n 3+ (A 41 ).n 4+ (A 61 ).n 6+ (A 71 ).n 7 - ( -(A 81 ).n 8 =0 

B 12 )-n 1 +(-B 21 -A 21 -B 2 3)-n 2 +(A3 2 +B3 2 )-n3+(A 52 )-n 5 - ( -(A 62 )-n 6 +(A 72 )-n 7 =0 
B 23 ) -n 2+ (- B 32 - A 32 - A 31 - A 34 ) -n 3 =0 

A 34) - n 3+ (" A 45" A 46" A 47" A 4l) •" 4=° 
A45)-n 4 +^A 5 2-A 58 j.n 5 =0 

A 46 )-n 4 +(-A 62 -A 61 )-n 6 =0 

A 47) -n 4 + (- A ?2 - A ?8 - A 71 ) -n 7=0 

A 58) n 5+ ( A 78) n 7+ (" A 8l) n 8=° 



n 1 + n2+n3+n 4 -ni 5 +-n 6 +n 7 +n 8 =l 



A solution set is found by the 
MinErr function. 



MinErr(n j,n 2 ,n 3,11 4 ,n 5 ,n 6 ,n 7 ,n g) = 



0.998 



1.5410 



r< 



2.03410 



1.49510 



.-7 



3.355-10 



4.289-10 



•v-7 



1.00210 
6.767-10"°. 



99 



The results are placed into vector, V, to be 
evaluated using the original set of equations. 



0.998 

1.54-10" 3 

2.034-10" 4 



V:= 



1.49510" 



3.35510 



,-7 



,-7 



4.289-10" 
1.00210" 
6.767-10" 1 



Assigns variables to the 
elements of vector V. 






o.o 



= V, 



1.0 



2.0 



3.0 



4.0 



= V, 



5.0 



: V, 



6.0 



V. 



7.0 



The variable b x represents the solved 
population fraction of level x. 

The positive terms are compared to the negative terms for each rate equation. 
Since the sum of all terms (for the first eight equations) equals zero, the sum of 
negative terms should equal the sum of positive terms. The exactness of the fit can be 
determined qualitatively through visual comparison of these sums for each equation. 

-B 12 )-bj=-5.71710 4 

B 21 + A 21 j.b 2+ (A 31 ).b3-,-(A 41 ).b 4+ (A 61 ).b 6+ (A 71 j.b 7+ !A 81i .b 8 = 5.717.10 4 



•B 21 - A 21 -B 23 )-b 2 : 



•6.09610 



12) - b 1 + (A 32+ B 32) " b 3+ ( A 52) " b 5+ ( A 62) >b 6+ ( A 72) " b 7 = 6 -°9610 4 



100 



-B 32 - A 32 - A 31 - A 34 ] b 3 =-3.80S10 3 
B 23 )-b 2 = 3.80410 3 



-A 45 - A 46 - A 4? - A 4 , l b 4 =- 13.569 
A 34)^3 = 13.567 



-A 52 -A 58 ).b 5 =-5.948 
A 45 ).b 4 = 5.95 



A 62- A 61- b 6=- 1 - 248 
A 46 )-b 4 = 1.248 



-A 72 - A 7g - A 71 )-b 7 =-0.112 
A 4 7)-b 4 =0.112 

-A 8 ,|b 8 =-1.956 
A 58)- b 5+( A 78)- b 7 = 1 - 956 



This last equation confirms that the sum of the population fractions 
is equal to one. 

b ]-fb 2 +b 3 + b 4 +b 5 + b 6 +b 7 +b g = 1.000 



REFERENCES 

1. H.R. Morris, C.C. Hoyt, C.C. Miller, P.J. Treado, Appl. Spectrosc, 50, 805 (1996). 

2. H. Coufal, U. Moller, S. Schneider, Appl. Optics, 21, 116 (1982). 

3. H.R. Morris, C.C. Hoyt, P.J. Treado, Appl. Spectrosc, 48, 857 (1993). 

4. A.F. Goetz, G. Vane, J.E. Solomon, B.N. Rock, Science, 4704, 1 147 (1985). 

5. O.I. Matveev, B.W. Smith, J.D. Winefordner, Appl. Phys. Lett., 11, 1673 (1998). 

6. A.A. Podshivalov, W.L. Clevenger, O.I. Matveev, B.W. Smith, J.D. Winefordner, 
Appl. Spectrosc, 54, 175 (2000). 

7. D. Pappas, O.I. Matveev, B.W. Smith, M.R. Shepard, A.A. Podshivalov, J.D. 
Winefordner, Appl. Optics, 39, 491 1 (2000). 

8. N.D. Finkelstein, W.R. Lempert, R.B. Miles, 19 th AIAA Advanced Measurement and 
Ground Testing Technology Conference, New Orleans, LA (1996). 

9. N.D. Finkelstein, W.R. Lempert, R.B. Miles, Optics Lett., 22, 1946 (1997). 

10. O.I. Matveev, B.W. Smith, J.D. Winefordner, Optics Lett., 23, 304 (1998). 

1 1. E. Korevaar, M. Rivers, C.S. Liu, SPIE Space Sensing, Communications, and 
Networking, 1059, 111 (1989). 

12. D. Pappas, N.C. Pixley, O.I. Matveev, B.W. Smith, J.D. Winefordner, Optics 
Commun., 191, 263 (2001). 

13. D. Pappas, N.C. Pixley, B.W. Smith, J.D. Winefordner, Spectrochim. Acta Part B, 56, 
1761 (2001). 

14. O.I. Matveev, B.W. Smith, J.D. Winefordner, Optics Commun., 156, 259 (1998). 

15. V.S. Letokhov, Laser Photoionization Spectroscopy, Academic Press, New York 
(1987). 

16. B.M. Patterson, T. Takekoshi, R.J. Knize, Phys. Rev. A, 59, 2508 (1999). 

17. K. Gerwert, K.J. Kollath,./. Phys. B: At. Mol. Phys., 16, L217 (1983). 



101 



102 



18. A.Z. Msezane, S.T. Manson, Phys. Rev. A, 29, 1594 (1984). 

19. C.E. Theodosiou, Phys. Rev. A, 33, 4400 (1986). 

20. O.I. Matveev, B.W. Smith, J.D. Winefordner, Appl. Optics, 36, 8833 (1997). 
21.0. Axner, P. Ljungberg, Spectrochim. Acta Rev., 15, 181 (1993). 

22. J.W. Daily, Appl. Optics, 16, 2322 (1977). 

23. J.B. Simeonsson, B.W. Smith, J.D. Winefordner, N. Omenetto, Appl. Spectrosc, 
45,521 (1991). 

24. G. Zizak, J.D. Bradshaw, J.D. Winefordner, Appl. Optics, 19, 3631 (1980). 

25. E. Voigtman, Spectrochim. Acta Part B, 48, 625 (1993). 

26. J.C. Travis, G.C. Turk (ed.), Laser-Enhanced Ionization Spectrometry, John Wiley & 
Sons, Inc., New York (1996). 

27. B. Warner, Mon. Not. R. Astr. Soc, 139, 115 (1968). 

28. J.J. More, B.S. Garbow, K.E. Hillstrom, Argonne National Laboratory publication 
number ANL-80-74 (1980). 

29. D. Boudreau, P. Ljungberg, O. Axner, Spectrochim. Acta Part B, 51, 413 (1996). 

30. D. Pappas, N.C. Pixley, B.W. Smith, J.D. Winefordner, Spectrochim. Acta Part 
5,55, 1503(2000). 

31. N.C. Pixley, T.L. Correll, D. Pappas, O.I. Matveev, B.W. Smith, J.D. Winefordner, 
Optics Lett, 26, 1946 (2001). 

32. N.C. Pixley, T.L. Correll, D. Pappas, B.W. Smith, J.D. Winefordner, Appl. 
Spectrosc, 56, 677 (2002). 

33. K. Shimoda (ed.), High Resolution Laser Spectroscopy, Springer-Verlag, New York 
(1976). 

34. E. Arimondo, M. Inguscio, P. Violino, Rev. Mod. Phys., 49, 31 (1977). 

35. S. Briandeau, D. Bloch, M. Ducloy, Phys Rev. A., 59, 3723 (1999). 

36. M. Tachikawa, K. Fukuda, S. Hayashi, T. Kawamura, Jpn. J. Appl. Phys., 37, 1556 
(1998). 



103 



37. S.E. Park, H.S. Lee, T.Y. Kwon, H. Cho, Optics Commun., 192, 49 (2001). 

38. O. Schmidt, K.M. Knaak, R. Wynands, D. Meschede, Appl. Physics B, 59, 167 
(1994). 

39. 1. Hirano, Optics Commun., 57, 331 (1986). 

40. P.P. Herrmann, J. Hoffnagle, A. Pedroni, N. Schlumpf, A. Weis, Optics Commun., 
56,22(1985). 

41. M. Samimy, M.P. Wernet, AIAA Journal, 38, 553 (2000). 

42. S.H. Bloom, R. Kremer, P.A. Searcy, M. Rivers, J. Menders, E. Korevaar, Optics 
Lett., 16, 1794(1991). 

43. S.H. Bloom, P.A. Searcy, K. Choi, R. Kremer, E. Korevaar, Optics Lett., 18, 244 
(1993). 

44. J.N. Forkey, N.D. Finkelstein, W.R. Lempert, R.B. Miles, AIAA Journal, 34, 442 
(1996). 

45. J.H. Grinstead, N.D. Finkelstein, W.R. Lempert, Optics Lett., 22, 331 (1997). 

46. O.I. Matveev, B.W. Smith, N. Omenetto, J.D. Winefordner, Spectrochim. Acta Part 
5,51,563(1996). 

47. W.L. Clevenger, O.I. Matveev, S. Cabredo, N. Omenetto, B.W. Smith, J.D. 
Winefordner, Anal. Chem., 69, 2237 (1997). 

48. F. Merkt, R.N. Zare, J. Chem. Phys., 101, 3495 (1994). 

49. T. Whitaker, G. Bekov, Spectrochim. Acta Part B., 49, 1557 (1994). 

50. S.E. Park, H.S. Lee, T.Y. Kwon, H. Cho, Optics Commun., 192, 57 (2001). 

51. D. Sarkisyan, D. Bloch, A. Papoyan, M. Ducloy, Optics Commun., 200, 201 (2001). 

52. M. Tachikawa, K. Fukuda, S. Hayashi, T. Kawamura, Jpn. J. Appl. Phys 37 LI 556 
(1998). 

53. S. Briaudeau, D. Bloch, M. Ducloy, Phys. Rev. A., 59, 3723 (1999). 

54. S. Briaudeau, S. Saltiel, G. Nienhuis, D. Bloch, M. Ducloy, Phys Rev A 57 3169 
(1998). ' ' ' 



104 



55. N. Omenetto (ed.), Analytical Laser Spectroscopy, John Wiley & Sons, New York 
(1979). 

56. K.J. Boiler, A. Imamoglu, S.E. Harris, Phys. Rev. Lett., 66, 2593 (1991). 

57. J.J. Clarke, W.A. Wijngaarden, Phys. Rev. A, 64, 64 (2001). 

58. J.J. Clarke, H. Chen, W. Wijngaarden, Appl. Optics, 40, 2047 (2001). 

59. D.J. Fulton, S. Shepard, R.R. Moseley, B.D. Sinclair, M.H. Dunn, Phys. Rev. A, 52, 
2302 (1995). 

60. H. Wang, D.J. Goorskey, W.H. Burkett, M. Xiao, Optics Lett., 25, 1732 (2000). 

61. N. Leinfellner, A. Huss, S. Baluschev, C. Neureiter, I.E. Mazets, E.A. Korsunsky, L. 
Windholz, Physica Scripta, 58, 583 (1998). 

62. S. Brandt, A. Nagel, R. Wynands, D. Mschede, Phys. Rev. A, 56, 1063 (1997). 



BIOGRAPHICAL SKETCH 
Nathan Charles Pixley was born in Huntingburg, Indiana, on January 30, 1976. 
He spent his early years in southern Illinois and Indiana. His parents, Charles and Peggy 
Pixley, had two additional children, Luke and Kristen, in 1978 and 1980, respectively. 
Nathan graduated from Castle High School in Newburgh, Indiana, in 1994. He then 
attended the University of Evansville in Evansville, Indiana, where he received a 
Bachelor of Science degree with a major in chemistry in the spring of 1998. He enrolled 
in the University of Florida in August 1998. 



105 



I certify that I have read this study and that in my opinion it conforms to acceptable 
standards of scholarly presentation and is fully adequate, in scope and quality, as a 
dissertation for the degree of Doctor of Philosophy. 






James D. WinefordneiyChairman 
'Graduate Research Professor of 
Chemistry 



I certify that I have read this study and that in my opinion it conforms to acceptable 
standards of scholarly presentation and is fully adequate, in scope and quality as a 
dissertation for the degree of Doctor of Philosophy. 




ichard A. Yost 
Professor of Chemistry 




I certify that I have read this study and that in my opinion it conforms to acceptable 
standards of scholarly presentation and is fully adequate, in scope and quality as a 
dissertation for the degree of Doctor of Philosophy. 




David H. Powell 
Scientist in Chemistry 



I certify that I have read this study and that in my opinion it conforms to acceptable 
standards of scholarly presentation and is fully adequate, in scope and quality, as a 
dissertation for the degree of Doctor of Philosophy. 




£ - §£ 



. Eyler 
sor of Chemistry 



I certify that I have read this study and that in my opinion it conforms to acceptable 
standards of scholarly presentation and is fully adequate, in scope and quality, as a 
dissertation for the degree of Doctor of Philosophy. 



-0 



nwav v 



Paul H. Holloway 
Professor of Materials Science and 
Engineering 



This dissertation was submitted to the Graduate Faculty of the Department of 
Chemistry in the College of Liberal Arts and Sciences and to the Graduate School and was 
accepted as a dissertation for the degree of Doctor of Philosophy. 



August 2002 



Dean, Graduate School 






UNIVERSITY OF FLORIDA 

lllllllllli 

3 1262 08555 3427